Research ArticleThe Identification of Marketing Performance Using TextMining of Airline Review Data

Jae-Won Hong1 and Seung-Bae Park 2

1Department of Global Trade Gyeongnam National University of Science and Technology 33 Dongjin-ro Jinju-siGyeongnam 52725 Republic of Korea2Department of Industrial Management Seoil University 28 Youngmasan-ro 90-gil Jungnang-gu Seoul Republic of Korea

Correspondence should be addressed to Seung-Bae Park sbparkseoilackr

Received 1 October 2018 Accepted 13 November 2018 Published 2 January 2019

Guest Editor Jaegeol Yim

Copyright copy 2019 Jae-Won Hong and Seung-Bae Park is is an open access article distributed under the Creative CommonsAttribution License which permits unrestricted use distribution and reproduction in anymedium provided the original work isproperly cited

We are aim firstly to extract major keywords using text mining method secondly to identify prominent keyword from thekeywords extracted from text mining analysis and then to confirm differences in influences of the keywords which affectcorporate performance Results were as following First keywords have been found to show distinctive features Since thekeywords posted from the clients showed certain tendency airlines accordingly need service management by identifying theservice property through keyword analysis Second prominent keywords have been found out of the keyword extracted from textmining Some of the keywords have significantly correlated with marketing performance but others not is implies that thecompany could uncover consumersrsquo needs through the prominent keywords and managing the properties related to theprominent keywords would help with improving corporate performanceird ldquorecommendrdquo should be treated distinctively withldquosatisfactionrdquo in terms of service management through the keywords Results suggest strategic implications to the practicalbusiness environment by analyzing keywords around the industry using text miningWe believe this work which aims to establishcommon ground for understanding these analyses across multiple disciplinary perspectives will encourage further research anddevelopment of service industry

1 Introduction

With rapid development of Internet environment and mobiledevices there have been inevitable changes in way of con-sumersrsquo communication Consumers form their own com-munity connected with SNS which refers to socialnetworking service such as Facebook Twitter blog etc Andthose network services are treated as reliable message sourcesover the mass media As social networks have become in-creasingly influential and consumers exploit much WOMinformation through the social network the information ofonline communication from social network plays a criticalrole in purchase decision making [1] Moreover when it issaid that traditional WOM communication is implicit orpersonal online WOM is more open and collective Due tothe characteristics of the social network reach of the in-formation is more extensive and the speed of the information

is also even faster Importance of active responses to con-sumersrsquo onlineWOM therefore increases in government andsocial organization as well as corporations

In this context online review that consumers produce onSNS can be influential means to grasp a tendency of con-sumers WOM and recommendation Consumers for in-stance even read product review to buy a book whichbelongs to the section of experiential products Precedingresearches on online review data mainly hire survey methodor text mining method [2] Survey method is used since itgets direct and clear opinions from consumers but is limitedto collect a variety of consumersrsquo opinions and latentopinions On the other hand using the text mining methodthe limitation of survey method can be overcome by col-lecting practical opinions from consumers and throughrefining data sourced from online it can also help withdeducing consumersrsquo underlying latent opinions as well

HindawiMobile Information SystemsVolume 2019 Article ID 1790429 8 pageshttpsdoiorg10115520191790429

Text mining on the other hand is a method of extractingunknown and valuable information from randomly orga-nized text data [3] us it is described as an automated toolthat extracts undisclosed information from text data whichare of unstructured format such as mail reviews webdocuments video clips or images [4] Recently majorstatistical packages and data mining programs include textmining function to facilitate or simplify the analysis processof those kinds of unstructured data ey provide withfunction of preprocess to conduct text mining and functionof summarizing and categorizing to identify the pattern ofthe data but also provides with a variety analysis function ofassociate analysis cluster analysis etc

It is important to reduce data in handling big data Largedata are not necessarily required to discover importantimplications e important thing is the appropriate datae academic significance of our research is to derive ways toreduce to appropriate data Current text mining methodtends to complement these limitations accordingly useful todeducemanagerial implication on practical decision-making[5] In this context by firstly deducing major variables basedon the keywords extracted by text mining to reduce largedata that is not necessary to discover important implicationand secondly combining them with items in questionnairethis research will help with suggesting practical implicationin the context of aviation industry

2 Related Works

21 Text Mining Text mining refers to automated methodsthat extract undiscovered and valuable information fromunstructured text by categorizing or structuralizing the text[6] By extracting information from big data in a variety offield connectivity within information will be uncoveredis overcomes limitations occurred by simple data analysisand enables to identify underlying meanings from massivetext data In this point the importance of these methodsincreases in terms of that the method can be utilized forsuggesting practical future strategies

rough the method of text mining researchers wouldtake advantage of not only extracting concepts of the textbut also identifying relationship with other concepts andvisualizing the relationship among the concepts Currentcontent analysis relies on items that researchers have ar-bitrarily selected accordingly extensive analysis on gathereddata therefore is limited and also external validity is notsecured since it relies on coders of the data Text mininghowever has been considered to surpass the limitation oftraditional content analysis and used in a variety of fieldsusing big data analysis social network analysis consumerproduct review analysis and other useful methods In otherwords text mining extracts the appropriate variables to limitthe breakdown of content analysis Netzer et al [7] examinedthe relationship between automobile brands through textmining and analyzed the market structure using the mul-tidimensional scaling method In addition Mostafa [8] hasclassified the lexicon through 3D Map in the research thatconfirmed the brand sentiments of famous brands such asNokia IBM and DHL through social network text mining

Text consists of words and analyzing text can be de-scribed as analyzing relationship among the words In termsof it text network analysis is also called semantic networkanalysis at is to say depending on the research it can becalled networks of words network text analysis semanticnets networks of concepts networks of centering words textnetwork analysis or semantic networks [9]

Text network analysis as mentioned above comple-ments the limitation of traditional content analysis andextracts underlying meaning that the text delivers More-over the pattern of text can be structurally analyzed toidentify the relationship among the meanings and the re-lationship accordingly can be visualized through the analysis[10] Text mining process has two phases including the dataprocess phase and data analysis phase e data processphase is relevant to data gathering and preprocess while dataanalysis phase is relevant to text analysis that extracts sig-nificant information from the text and visualizing in-formation and extracting knowledge from the formeranalysis [6]

rough this process large-volume data can be mademore data-suitable for analysis enabling continuous re-search to compare experiments

22 Consumersrsquo Evaluation Criteria on Airlines Aviationindustry can be described as a field where degree of in-teraction customization and labour intensiveness is rela-tively lower than that of other field [11] Service comparedto products has shown distinctive features of intangibilityheterogeneity inseparability and perishability [12] In otherwords consumers cannot see or touch purchased ldquoservicesrdquoIn addition production and consumption take place at thesame time and it cannot be stored but even extinguishesonce it is unused after production In that properties ofservice therefore perceived service evaluation or recom-mendation can be crucial indicators in the field of aviationindustry Jia [13] has conducted text mining though aChinese crowd-sourced online review community and49080 pairs of restaurant ratings and reviews were exam-ined with high-frequency words major topics and sub-topics identified After text mining multilinear regressionwas employed to screen out the most impactful factors thatinfluence taste environment and service ratings Manage-rially the idea of triggering the synergistic benefit fromcustomer ratings and reviews is referential for marketpractitioners both within and beyond the catering industry

In case of aviation industry these elements includingflight schedule fare services punctuality comfort of seatssafety and frequent flyer program can be determinants ofservice satisfaction on airlines (IATA International AirTransport Association) And in research of Hong and Park[14] factors that determine consumersrsquo satisfaction onairline services include punctuality safety courteousagents clean equipment space desirable schedule prof-itability of the airline reliability on the provided serviceand financial costs In addition flight fare boardingprocess space and comfort of seats in-flight meal baggagedelivery and ticketing process also can affect consumersrsquo

2 Mobile Information Systems

satisfaction Based on researches passengers who experi-ence airline services tend to compare and evaluate itsoverall quality on the basis of tangible and intangibleservices provided by airlines Will these factors be derivedfrom the customersrsquo texts in Internet Internet commentsbased on anonymity reveal the desire of consumers andreveal more clear facts

Aviation industry itself can be defined as a service in-dustry where degree of interaction customization and labourintensiveness is relatively lower than that of other field [15]In-flight service which is the interest of this study howevercan be described with high interaction high customizationand high labour intensiveness us assessment and rec-ommendation on services which consumers directly haveexperienced will be important indicator for the industry Ingeneral in-flight service consists of tangible service andhuman service that help with passengersrsquo travel experiencescomfort of seats in-flight employee services in-flight foodand beverage entertaining services etc ese elementswhich are important in the aviation industry are identifiedthrough text mining and classified through cluster analysisese studies provide a possibility to become more realisticstudies by combining text mining and statistical analysis

3 Methodologies

31 Research Process In this study we conducted the fol-lowing process to find out whether the core keywords can beidentified through text mining and whether the core key-words are important for the performance of a company (seeFigure 1) First we gathered online reviews of customers andextracted core keywords from it by text mining methodAnd we conducted text clustering analysis to explore themeaning of extracted core keywords After that we con-ducted empirical test for demonstrating the impact ofcore keywords through analysis of the relevance of thecompanyrsquos marketing performance such as satisfactionrecommendation

32 Data is study used the online review data of airlinecustomers which was provided by global air service eval-uation agency Skytrax in United Kingdom is study settwo large air carriers in Korea and Japan e main contentsof the data consist of two types of data One type is the textdata containing the customerrsquos experience after using the airservice e other type is the survey data which include theevaluation of services satisfaction and recommendationafter using the air service e questionnaire survey wasconducted online for customers who have recently usedairline e respondents are confirmed by presenting theirplane ticket e item of customer satisfaction was measuredas 10 point-Likert scale and recommendation intention wasmeasured as binomial scale

Data period is 3 years from January 2013 to December2015 In the data period 197 reviews were for the Koreancarrier and 214 reviews for Japanese carrier So this studycollected the review data on 411 people in total during dataperiod and used them for analysis Among the collected data

text data were used for core keyword extraction andquestionnaire data were used for empirical analysis to verifythe influence of core keywords

e number of countries to which the respondents wereaffiliated was 32 in total and the United States was found tobe the most (338) followed by Australia (127) UnitedKingdom (107) Canada (61) and Singapore (51)Table 1 shows the top 10 countries on the basis of thenumber of respondents

33 Analytic Process In order to analyze the text data thisstudy performed two processes Text mining process was forextracting words and text clustering process was for ana-lyzing meaning based on extracted words

e text mining process was a process of extractingwords from a text document and creating word data Andthis process included a step of extracting meaningfulwords based on text data e detailed procedure is asfollows [6]

331 Text Mining Process

Step 1 (treat text) is is the process of creating word databased on words contained in a text documents

tdrarr

tf d t1( 1113857( 1113857 tf d t2( 1113857( 1113857 tf d tm( 1113857( 1113857 (1)

where tdrarr

is the words vector D is the documents setd1 d2 dn1113864 1113865 and T is the words set t1 t2 tm1113864 1113865

Step 2 (extract words) is is the process of extractingmeaningful words by assigning weights through the ap-pearance frequency of words between documents set

W(t) 1 +1

log2|D|1113944disinD

P(d t) log2 P(d t)

with P(d t) tf(d t)

1113936nl1tf dl t( 1113857

(2)

where tf(d t) is the frequency of word t in a document d

332 Text Clustering Process On the other hand textclustering process is a process of clustering through thedistance between words e cluster is set up and adjusted

Text clustering

Empirical analysis

Text mining

Data collection bull Online review data

bull Keywords extraction

bull Keywords classification

bull Keywords effect test

Figure 1 Research process

Mobile Information Systems 3

after the cluster was set In this method (1) generate k initialseed randomly within data domain (2) create k clusters byassociating every observation with the nearest seed positionand (3) adjust the center position of each of the k clustersusing the average of the observations belonging to thecluster By repeating this process until all observations areassociated it can make clusters which have similar obser-vations corresponding to the number of k

Step 1 (cluster setting) Set up a cluster with random seeds inthe data and the cluster is grouped by the Euclidean distancebetween each data and seed based on near distance

min S 1113944k

i11113944xisinsi

xminus ui

2

(3)

where x is the words set x1 x2 xn1113864 1113865 s is the clusters sets1 s2 sk1113864 1113865 and ui is the average of cluster si

Step 2 (cluster adjust) Reset the value of cluster and adjustthe position of cluster by using the average of the data in thecluster

4 Results

41 Keywords Extraction by Text Mining In this studyseveral preprocessing steps were performed to improve theaccuracy of text mining

First all uppercase letters in text data are converted tolowercase letters to unify words Also we removed un-necessary special characters such as ldquordquo ldquordquo and so on Andthe definite article (ldquotherdquo) the indefinite article (ldquoardquo ldquoanrdquo)prepositions (ldquoofrdquo ldquoinrdquo ldquoforrdquo ldquothroughrdquo etc) and pronouns(ldquoitrdquo ldquotheirrdquo ldquohisrdquo etc) were also removed rough thisprocess a total of 3774 keywords were extracted e key-words were rearranged based on the sparsity which representsthe ratio of the whitespace in the matrix because analyzing allkeywords makes it difficult to explain meaningful results

ere were 11 keywords based on the sparsity 08 19keywords based on the sparsity 085 45 keywords based onthe sparsity 09 and 132 keywords based on the sparsity 095In this study a total of 45 keywords were extracted byapplying the sparsity 09 and it is used for analysis Too few

keywords are insufficient to understand customer behaviourbut too many keywords can distort results with less im-portant words In this analysis 45 keywords were selectedconsidering the usefulness of the interpretation whilereflecting the customer behaviour

e keywords with high frequency were ldquogoodrdquo ldquoseatsrdquoldquocabinrdquo ldquoclassrdquo ldquoexcellentrdquo ldquocomfortablerdquo ldquostaffrdquo and soon Figure 2 shows the frequency distribution of thesekeywords and Figure 3 shows the frequency of keywordappearance by the word cloud method

42 Keywords Classification by Text Clustering In this studywe performed keyword classification analysis for semanticanalysis of extracted 45 keywords e results are as followsKeyword classification was based on hierarchical clusteringand the distance between the keywords was measured by theEuclidean method

As a result of clustering analysis 45 keywords wereclassified into two clusters Cluster 1 consisted of servicecontent such as ldquoseatsrdquo ldquocabinrdquo ldquoclassrdquo and ldquostaffrdquo whoprovide the service And it has service evaluation-relatedkeywords such as ldquogoodrdquo ldquoexcellentrdquo ldquocomfortablerdquo etc

On the other hand Cluster 2 consisted of more detailservice content such as ldquomealrdquo ldquomovierdquo ldquodrinksrdquo ldquocheckrdquoldquoservedrdquo and so on Also it has service evaluation-relatedkeywords such as ldquogreatrdquo ldquonicerdquo ldquowellrdquo ldquobestrdquo ldquocleanrdquo etc(see Figure 4)

Figure 5 shows the result of keyword visualization withtwo clusters based on the k-means method for better insighte keywords in Cluster 1 are formed to be spaced apartfrom each other and the keywords in Cluster 2 are formed tobe relatively wide spacing And the two componentsexplained 6183 of the point variability

43 Effect of Core Keywords In this study we performedcombining analysis with core keywords data extracted bytext mining and questionnaire respondent data to overcomethe disadvantages of existing text mining research and toprovide practical implications to the aviation industry Inother words since core keywords appear as a result of textmining and cluster analysis can be considered to representonline reviews of airline customers we attempted to un-derstand the meaning of representative keywords by ex-amining the influence of each evaluation concept oncustomer satisfaction and customer recommendation basedon the evaluation of these customers

In the analysis we analyzed the effect of the four key-words such as ldquoseatsrdquo ldquostaffrdquo ldquocabinrdquordquo and ldquoclassrdquo which arethe main keywords corresponding to the contents of theCluster 1 on customer satisfaction and recommendation ereason for this is that adjective and adverbial keywords amongthe keywords of Cluster 1 are mainly keywords indicating theresult of the service And the keywords appearing in Cluster 2are keywords that deal with details of the service contents ofCluster 1 So we define that these keywords are conceptuallyincluded in the above four top keywords

In the analytical model the questionnaire itemswere used as variables Independent variables were the

Table 1 Characteristics of country

No CountryKorea Japan Total

N N N 1 United States 58 294 81 379 139 3382 Australia 50 254 2 09 52 1273 United Kingdom 25 127 19 89 44 1074 Canada 4 20 21 98 25 615 Singapore 3 15 18 84 21 516 Japan 3 15 15 70 18 447 South Korea 13 66 1 05 14 348 Brazil 1 05 12 56 13 329 Germany 7 36 4 19 11 2710 New Zealand 8 41 1 05 9 22

4 Mobile Information Systems

questionnaire items for service evaluation on seats staffcabin and class And dependent variables were satisfac-tion and recommendation after using airline service

As a result of analysis seats staff and cabin were foundto be core keywords which had a significant effect on

satisfaction but class did not (see Table 2) In other wordsthe WOM (word of mouth) of the customers related to seatand staff services influence directly on customer satisfactionOn the other hand the customers of the airline talk fre-quently about class type such as first class and business classbut the class does not affect the customer satisfaction

is means that the airline should pay particular at-tention to the seat service and the staffrsquos efforts are needed tocomplement their service It also suggests that customersatisfaction can be improved by paying attention to theservice environment factors in the cabin

On the other hand seats staff cabin and class have asignificant effect on recommendation And in terms of powerof influence seats service was found to have a greater effect onrecommendation than staff serviceese results indicate thatseat service is also an important factor in customer recom-mendation and the service of airline crew can complement itHowever cabin and class service did not have a significanteffect on recommendation (Table 3) ese results indicatethat although airline passengers talk a lot about the services inthe cabin they do not really have much to do with customerreferrals on recommendation As a result it is suggested thatseats and staff services are especially required to provideairline with recommendations from their customers

5 Conclusion

Data mining techniques need to be used in the aviationindustry to understand consumer behaviour at is this

0

100

200

Word

Freq

Always

Atte

ntive

Best

Bette

rBit

Boarding

Busin

ess

Cabin

Check

Checkin

Choice

Class

Clean

Comfortable

Definitely

Drin

ksEcon

omy

Efficient

Entertainm

ent

Excellent

Experie

nce

Flat

Friend

lyFu

llGoo

dGreat

Groun

dHelpful

How

ever

Infligh

tLeg

Legs

Like

Limited

Loun

geMade

Many

Meal

Meals

Movies

Much

Nice

Offe

red

Onb

oard

Overall

Polite

Quality

Quite

Really

Screen

Seats

Selection

Served

Sleep

Space

Staff

Well

Figure 2 Distribution of keyword frequency

Figure 3 Word cloud of keywords

Mobile Information Systems 5

Great

Meal

Loun

geNice

Entertainm

ent

Infligh

tWell

Best

Experie

nce

Really

Meals

Served

Friend

lyHow

ever leg

Sleep

Overall

Quite bit

Bette

rMuch

Quality

Movies

Selection

Check

Made

Offered

Boarding

Clean

Onb

oard

Efficient

Helpful

Polite

Atte

ntive

Choice

Drin

ksGoo

dSeats

Staff

Comfortable

Excellent

Cabin

Econ

omy

Busin

ess

Class

5

10

15

20

25

30Heigh

t

Figure 4 Cluster dendrogram

ndash15 ndash10 ndash5 0 5

ndash4

ndash2

0

2

4

6

Component 1

Com

pone

nt 2

AttentiveBest Better

BitBoarding

Business

CabinCheck Choice

Class

CleanComfortable

DrinksEconomy

Efficient

EntertainmentExcellent

Experience

Friendly

Good

Great

HelpfulHoweverInflightLeg

Lounge

MadeMeal

Meals

Movies

Much

Nice

OfferedOnboardOverallPolite

Quality

QuiteReally

Seats SelectionServed Sleep

Staff

Well

1

2

Figure 5 Cluster plot ese two components explain 6183 of the point variability

Table 2 Effects of core keyword on satisfaction

Variables B β t-value p-valueSeats 0987 0513 186lowastlowastlowast 0001Staff 0874 0334 112lowastlowastlowast 0001Cabin 0555 0235 74lowastlowastlowast 0001Class 0056 0011 05 0645lowastlowastlowastρlt 0001 F value 3424 R2 0771

Table 3 Effects of core keyword on recommendation

Variables B β t-value p-valueSeats 0156 0536 135lowastlowastlowast 0001Staff 0110 0275 64lowastlowastlowast 0001Cabin 0020 0055 12 0224Class 0039 0051 15 0139lowastlowastlowastρlt 0001 F value 1110 R2 0518

6 Mobile Information Systems

study used text mining to develop a unified understanding ofkeywords in the aviation industry in a data-driven wayBased on the airline review data we proposed a two-stepprocess of extracting key keywords by text mining andgrouping them into cluster analysis Specifically we useda combination of metrics and clustering algorithms topreprocess and analyze text data related to keywords ex-traction method including text from the scientific literatureand news articles is study seeks firstly for identifyingprominent keywords at consumersrsquo side using text miningmethod on consumersrsquo online review data and then forconfirming influences that the keywords affect corporatemarketing performance is study is not only the researchthat searches for key keywords [16] but also the researchthat identifies marketing performance in the aviation in-dustry through text mining in service category level Con-clusion and implication are as follows

First keywords are shown to have distinctive clustercharacteristics As a result of identifying characteristics ofmajor keywords through clustering the keywords have beenclassified into three sections including service that airlinesprovides with details of each service and assessment on theservices In other words since the service provided by airlineand the details of each service are differentiated at differentlevels it indicates that service management should becentered on core service elements that can be clearly rec-ognized in order to improve service evaluation

Second the key keywords extracted from text miningwere found to have a different relationship with corporateperformance In other words the service of the seats and thestaff was more important to the company performance thanthe cabin or class ese results show that comfort is the keycustomerrsquos needs in long-distance air travel and that theservice focused on the comfort of the seat or the comfort ofthe seat is an important factor in corporate performance

ird recommendation and satisfaction should bemanaged distinctively in the service management of theaviation industry e results show that keywords that affectconsumersrsquo satisfaction and keywords that affect consumersrsquorecommendation are found to differ and the degree ofimpacts is also shown to be different as well In other wordsseats staff and cabin are important factors to improve thesatisfaction and recommendation of consumers Seats arethe most important factors in consumer satisfaction andrecommendation but relatively more staff and cabin areimportant for consumer satisfaction and seats are moreimportant for consumer recommendation ese resultsshow that the human service of the crew which can berelatively subjective is more influential in satisfaction Onthe other hand seats are more important for relativelyobjective recommendations

is study presents academic implication that the studyhas extended its application area of text mining It hascurrently focused on exploratory study while this study hasextended it to study field of cause and effect Moreover theresearch also presents practical implication for corporationsto efficiently manage keywords In spite of the implicationsand advantage of text mining [17] this study has limitationsFirst the research range is limited to two airlines in Korea

and Japan Secondly since the review data are producedspontaneously by clients collected data can be biased andlimited to active clients who are willing to express hisherexperience erefore in future study selecting more rep-resentative airlines is needed and also minimizing conve-nience of the respondents is needed to generalize the resultsof the study in the future

Data Availability

e text mining data used to support the findings of thisstudy are available from the authors upon request e dataused in this study are Airline Reviews and Rating data fromSkytrax (httpwwwairlinequalitycom)

Conflicts of Interest

e authors declare that they have no conflicts of interest

References

[1] P Nayak and A K Chatterjee ldquoEffects of aluminium exposureon brain glutamate and GABA systems an experimental studyin ratsrdquo Food and Chemical Toxicology vol 39 no 12pp 1285ndash1289 2001

[2] J A Chevalier and D Mayzlin ldquoe effect of word of mouthon sales online book reviewsrdquo Journal of Marketing Researchvol 43 no 3 pp 345ndash354 2018

[3] W Fan LWallace S Rich and Z Zhang ldquoTapping the powerof text miningrdquo Communications of the ACM vol 49 no 9pp 76ndash82 2006

[4] D Meyer K Hornik and I Feinerer ldquoText mining in-frastructure in Rrdquo Journal of Statistical Software vol 25 no 5pp 1ndash54 2008

[5] R V Kozinets K de Valck A CWojnicki and S J SWilnerldquoNetworked narratives understanding word-of-mouth mar-keting in online communitiesrdquo Journal of Marketing vol 74no 2 pp 71ndash89 2010

[6] A Hotho A Nurnberger and G Paaszlig ldquoA brief survey of textminingrdquo Ldv Forum vol 20 no 1 pp 19ndash62 2005

[7] O Netzer R Feldman J Goldenberg and M Fresko ldquoMineyour own business market-structure surveillance through textminingrdquo Marketing Science vol 31 no 3 pp 521ndash543 2012

[8] M M Mostafa ldquoMore than words social networksrsquo textmining for consumer brand sentimentsrdquo Expert Systems withApplications vol 40 no 10 pp 4241ndash4251 2013

[9] D Paranyushkin ldquoText network analysis (2010)rdquo in Con-ference du Performing Arts Forum 2011 httpnoduslabscomresearchpathways-meaning-circulation

[10] R Feldman and J Sanger gte Text Mining Handbook Ad-vanced Approaches in Analysing Un-Structured Data Cam-bridge University Press Cambridge UK 2007

[11] R W Schmenner ldquoHow can service businesses survive andprosperrdquo SloanManagement Review1986-1998 vol 27 no 3p 21 1986

[12] J Fitzsimmons and M Fitzsimmons Service managementOperations Strategy Information Technology McGraw-HillHigher Education New York City PA USA 2013

[13] S Jia ldquoBehind the ratings text mining of restaurant cus-tomersrsquo online reviewsrdquo International Journal of MarketResearch vol 60 no 6 pp 561ndash572 2018

[14] J-W Hong and S-B Park ldquoStudy on the extraction of corekeywords and its effects through text miningrdquo International

Mobile Information Systems 7

satisfaction Based on researches passengers who experi-ence airline services tend to compare and evaluate itsoverall quality on the basis of tangible and intangibleservices provided by airlines Will these factors be derivedfrom the customersrsquo texts in Internet Internet commentsbased on anonymity reveal the desire of consumers andreveal more clear facts

Aviation industry itself can be defined as a service in-dustry where degree of interaction customization and labourintensiveness is relatively lower than that of other field [15]In-flight service which is the interest of this study howevercan be described with high interaction high customizationand high labour intensiveness us assessment and rec-ommendation on services which consumers directly haveexperienced will be important indicator for the industry Ingeneral in-flight service consists of tangible service andhuman service that help with passengersrsquo travel experiencescomfort of seats in-flight employee services in-flight foodand beverage entertaining services etc ese elementswhich are important in the aviation industry are identifiedthrough text mining and classified through cluster analysisese studies provide a possibility to become more realisticstudies by combining text mining and statistical analysis

3 Methodologies

31 Research Process In this study we conducted the fol-lowing process to find out whether the core keywords can beidentified through text mining and whether the core key-words are important for the performance of a company (seeFigure 1) First we gathered online reviews of customers andextracted core keywords from it by text mining methodAnd we conducted text clustering analysis to explore themeaning of extracted core keywords After that we con-ducted empirical test for demonstrating the impact ofcore keywords through analysis of the relevance of thecompanyrsquos marketing performance such as satisfactionrecommendation

32 Data is study used the online review data of airlinecustomers which was provided by global air service eval-uation agency Skytrax in United Kingdom is study settwo large air carriers in Korea and Japan e main contentsof the data consist of two types of data One type is the textdata containing the customerrsquos experience after using the airservice e other type is the survey data which include theevaluation of services satisfaction and recommendationafter using the air service e questionnaire survey wasconducted online for customers who have recently usedairline e respondents are confirmed by presenting theirplane ticket e item of customer satisfaction was measuredas 10 point-Likert scale and recommendation intention wasmeasured as binomial scale

Data period is 3 years from January 2013 to December2015 In the data period 197 reviews were for the Koreancarrier and 214 reviews for Japanese carrier So this studycollected the review data on 411 people in total during dataperiod and used them for analysis Among the collected data

text data were used for core keyword extraction andquestionnaire data were used for empirical analysis to verifythe influence of core keywords

e number of countries to which the respondents wereaffiliated was 32 in total and the United States was found tobe the most (338) followed by Australia (127) UnitedKingdom (107) Canada (61) and Singapore (51)Table 1 shows the top 10 countries on the basis of thenumber of respondents

33 Analytic Process In order to analyze the text data thisstudy performed two processes Text mining process was forextracting words and text clustering process was for ana-lyzing meaning based on extracted words

e text mining process was a process of extractingwords from a text document and creating word data Andthis process included a step of extracting meaningfulwords based on text data e detailed procedure is asfollows [6]

331 Text Mining Process

Step 1 (treat text) is is the process of creating word databased on words contained in a text documents

tdrarr

tf d t1( 1113857( 1113857 tf d t2( 1113857( 1113857 tf d tm( 1113857( 1113857 (1)

where tdrarr

is the words vector D is the documents setd1 d2 dn1113864 1113865 and T is the words set t1 t2 tm1113864 1113865

Step 2 (extract words) is is the process of extractingmeaningful words by assigning weights through the ap-pearance frequency of words between documents set

W(t) 1 +1

log2|D|1113944disinD

P(d t) log2 P(d t)

with P(d t) tf(d t)

1113936nl1tf dl t( 1113857

(2)

where tf(d t) is the frequency of word t in a document d

332 Text Clustering Process On the other hand textclustering process is a process of clustering through thedistance between words e cluster is set up and adjusted

Text clustering

Empirical analysis

Text mining

Data collection bull Online review data

bull Keywords extraction

bull Keywords classification

bull Keywords effect test

Figure 1 Research process

Mobile Information Systems 3

after the cluster was set In this method (1) generate k initialseed randomly within data domain (2) create k clusters byassociating every observation with the nearest seed positionand (3) adjust the center position of each of the k clustersusing the average of the observations belonging to thecluster By repeating this process until all observations areassociated it can make clusters which have similar obser-vations corresponding to the number of k

Step 1 (cluster setting) Set up a cluster with random seeds inthe data and the cluster is grouped by the Euclidean distancebetween each data and seed based on near distance

min S 1113944k

i11113944xisinsi

xminus ui

2

(3)

where x is the words set x1 x2 xn1113864 1113865 s is the clusters sets1 s2 sk1113864 1113865 and ui is the average of cluster si

Step 2 (cluster adjust) Reset the value of cluster and adjustthe position of cluster by using the average of the data in thecluster

4 Results

41 Keywords Extraction by Text Mining In this studyseveral preprocessing steps were performed to improve theaccuracy of text mining

First all uppercase letters in text data are converted tolowercase letters to unify words Also we removed un-necessary special characters such as ldquordquo ldquordquo and so on Andthe definite article (ldquotherdquo) the indefinite article (ldquoardquo ldquoanrdquo)prepositions (ldquoofrdquo ldquoinrdquo ldquoforrdquo ldquothroughrdquo etc) and pronouns(ldquoitrdquo ldquotheirrdquo ldquohisrdquo etc) were also removed rough thisprocess a total of 3774 keywords were extracted e key-words were rearranged based on the sparsity which representsthe ratio of the whitespace in the matrix because analyzing allkeywords makes it difficult to explain meaningful results

ere were 11 keywords based on the sparsity 08 19keywords based on the sparsity 085 45 keywords based onthe sparsity 09 and 132 keywords based on the sparsity 095In this study a total of 45 keywords were extracted byapplying the sparsity 09 and it is used for analysis Too few

keywords are insufficient to understand customer behaviourbut too many keywords can distort results with less im-portant words In this analysis 45 keywords were selectedconsidering the usefulness of the interpretation whilereflecting the customer behaviour

e keywords with high frequency were ldquogoodrdquo ldquoseatsrdquoldquocabinrdquo ldquoclassrdquo ldquoexcellentrdquo ldquocomfortablerdquo ldquostaffrdquo and soon Figure 2 shows the frequency distribution of thesekeywords and Figure 3 shows the frequency of keywordappearance by the word cloud method

42 Keywords Classification by Text Clustering In this studywe performed keyword classification analysis for semanticanalysis of extracted 45 keywords e results are as followsKeyword classification was based on hierarchical clusteringand the distance between the keywords was measured by theEuclidean method

As a result of clustering analysis 45 keywords wereclassified into two clusters Cluster 1 consisted of servicecontent such as ldquoseatsrdquo ldquocabinrdquo ldquoclassrdquo and ldquostaffrdquo whoprovide the service And it has service evaluation-relatedkeywords such as ldquogoodrdquo ldquoexcellentrdquo ldquocomfortablerdquo etc

On the other hand Cluster 2 consisted of more detailservice content such as ldquomealrdquo ldquomovierdquo ldquodrinksrdquo ldquocheckrdquoldquoservedrdquo and so on Also it has service evaluation-relatedkeywords such as ldquogreatrdquo ldquonicerdquo ldquowellrdquo ldquobestrdquo ldquocleanrdquo etc(see Figure 4)

Figure 5 shows the result of keyword visualization withtwo clusters based on the k-means method for better insighte keywords in Cluster 1 are formed to be spaced apartfrom each other and the keywords in Cluster 2 are formed tobe relatively wide spacing And the two componentsexplained 6183 of the point variability

43 Effect of Core Keywords In this study we performedcombining analysis with core keywords data extracted bytext mining and questionnaire respondent data to overcomethe disadvantages of existing text mining research and toprovide practical implications to the aviation industry Inother words since core keywords appear as a result of textmining and cluster analysis can be considered to representonline reviews of airline customers we attempted to un-derstand the meaning of representative keywords by ex-amining the influence of each evaluation concept oncustomer satisfaction and customer recommendation basedon the evaluation of these customers

In the analysis we analyzed the effect of the four key-words such as ldquoseatsrdquo ldquostaffrdquo ldquocabinrdquordquo and ldquoclassrdquo which arethe main keywords corresponding to the contents of theCluster 1 on customer satisfaction and recommendation ereason for this is that adjective and adverbial keywords amongthe keywords of Cluster 1 are mainly keywords indicating theresult of the service And the keywords appearing in Cluster 2are keywords that deal with details of the service contents ofCluster 1 So we define that these keywords are conceptuallyincluded in the above four top keywords

In the analytical model the questionnaire itemswere used as variables Independent variables were the

Table 1 Characteristics of country

No CountryKorea Japan Total

N N N 1 United States 58 294 81 379 139 3382 Australia 50 254 2 09 52 1273 United Kingdom 25 127 19 89 44 1074 Canada 4 20 21 98 25 615 Singapore 3 15 18 84 21 516 Japan 3 15 15 70 18 447 South Korea 13 66 1 05 14 348 Brazil 1 05 12 56 13 329 Germany 7 36 4 19 11 2710 New Zealand 8 41 1 05 9 22

4 Mobile Information Systems

questionnaire items for service evaluation on seats staffcabin and class And dependent variables were satisfac-tion and recommendation after using airline service

As a result of analysis seats staff and cabin were foundto be core keywords which had a significant effect on

satisfaction but class did not (see Table 2) In other wordsthe WOM (word of mouth) of the customers related to seatand staff services influence directly on customer satisfactionOn the other hand the customers of the airline talk fre-quently about class type such as first class and business classbut the class does not affect the customer satisfaction

is means that the airline should pay particular at-tention to the seat service and the staffrsquos efforts are needed tocomplement their service It also suggests that customersatisfaction can be improved by paying attention to theservice environment factors in the cabin

On the other hand seats staff cabin and class have asignificant effect on recommendation And in terms of powerof influence seats service was found to have a greater effect onrecommendation than staff serviceese results indicate thatseat service is also an important factor in customer recom-mendation and the service of airline crew can complement itHowever cabin and class service did not have a significanteffect on recommendation (Table 3) ese results indicatethat although airline passengers talk a lot about the services inthe cabin they do not really have much to do with customerreferrals on recommendation As a result it is suggested thatseats and staff services are especially required to provideairline with recommendations from their customers

5 Conclusion

Data mining techniques need to be used in the aviationindustry to understand consumer behaviour at is this

0

100

200

Word

Freq

Always

Atte

ntive

Best

Bette

rBit

Boarding

Busin

ess

Cabin

Check

Checkin

Choice

Class

Clean

Comfortable

Definitely

Drin

ksEcon

omy

Efficient

Entertainm

ent

Excellent

Experie

nce

Flat

Friend

lyFu

llGoo

dGreat

Groun

dHelpful

How

ever

Infligh

tLeg

Legs

Like

Limited

Loun

geMade

Many

Meal

Meals

Movies

Much

Nice

Offe

red

Onb

oard

Overall

Polite

Quality

Quite

Really

Screen

Seats

Selection

Served

Sleep

Space

Staff

Well

Figure 2 Distribution of keyword frequency

Figure 3 Word cloud of keywords

Mobile Information Systems 5

Great

Meal

Loun

geNice

Entertainm

ent

Infligh

tWell

Best

Experie

nce

Really

Meals

Served

Friend

lyHow

ever leg

Sleep

Overall

Quite bit

Bette

rMuch

Quality

Movies

Selection

Check

Made

Offered

Boarding

Clean

Onb

oard

Efficient

Helpful

Polite

Atte

ntive

Choice

Drin

ksGoo

dSeats

Staff

Comfortable

Excellent

Cabin

Econ

omy

Busin

ess

Class

5

10

15

20

25

30Heigh

t

Figure 4 Cluster dendrogram

ndash15 ndash10 ndash5 0 5

ndash4

ndash2

0

2

4

6

Component 1

Com

pone

nt 2

AttentiveBest Better

BitBoarding

Business

CabinCheck Choice

Class

CleanComfortable

DrinksEconomy

Efficient

EntertainmentExcellent

Experience

Friendly

Good

Great

HelpfulHoweverInflightLeg

Lounge

MadeMeal

Meals

Movies

Much

Nice

OfferedOnboardOverallPolite

Quality

QuiteReally

Seats SelectionServed Sleep

Staff

Well

1

2

Figure 5 Cluster plot ese two components explain 6183 of the point variability

Table 2 Effects of core keyword on satisfaction

Variables B β t-value p-valueSeats 0987 0513 186lowastlowastlowast 0001Staff 0874 0334 112lowastlowastlowast 0001Cabin 0555 0235 74lowastlowastlowast 0001Class 0056 0011 05 0645lowastlowastlowastρlt 0001 F value 3424 R2 0771

Table 3 Effects of core keyword on recommendation

Variables B β t-value p-valueSeats 0156 0536 135lowastlowastlowast 0001Staff 0110 0275 64lowastlowastlowast 0001Cabin 0020 0055 12 0224Class 0039 0051 15 0139lowastlowastlowastρlt 0001 F value 1110 R2 0518

6 Mobile Information Systems

study used text mining to develop a unified understanding ofkeywords in the aviation industry in a data-driven wayBased on the airline review data we proposed a two-stepprocess of extracting key keywords by text mining andgrouping them into cluster analysis Specifically we useda combination of metrics and clustering algorithms topreprocess and analyze text data related to keywords ex-traction method including text from the scientific literatureand news articles is study seeks firstly for identifyingprominent keywords at consumersrsquo side using text miningmethod on consumersrsquo online review data and then forconfirming influences that the keywords affect corporatemarketing performance is study is not only the researchthat searches for key keywords [16] but also the researchthat identifies marketing performance in the aviation in-dustry through text mining in service category level Con-clusion and implication are as follows

First keywords are shown to have distinctive clustercharacteristics As a result of identifying characteristics ofmajor keywords through clustering the keywords have beenclassified into three sections including service that airlinesprovides with details of each service and assessment on theservices In other words since the service provided by airlineand the details of each service are differentiated at differentlevels it indicates that service management should becentered on core service elements that can be clearly rec-ognized in order to improve service evaluation

Second the key keywords extracted from text miningwere found to have a different relationship with corporateperformance In other words the service of the seats and thestaff was more important to the company performance thanthe cabin or class ese results show that comfort is the keycustomerrsquos needs in long-distance air travel and that theservice focused on the comfort of the seat or the comfort ofthe seat is an important factor in corporate performance

ird recommendation and satisfaction should bemanaged distinctively in the service management of theaviation industry e results show that keywords that affectconsumersrsquo satisfaction and keywords that affect consumersrsquorecommendation are found to differ and the degree ofimpacts is also shown to be different as well In other wordsseats staff and cabin are important factors to improve thesatisfaction and recommendation of consumers Seats arethe most important factors in consumer satisfaction andrecommendation but relatively more staff and cabin areimportant for consumer satisfaction and seats are moreimportant for consumer recommendation ese resultsshow that the human service of the crew which can berelatively subjective is more influential in satisfaction Onthe other hand seats are more important for relativelyobjective recommendations

is study presents academic implication that the studyhas extended its application area of text mining It hascurrently focused on exploratory study while this study hasextended it to study field of cause and effect Moreover theresearch also presents practical implication for corporationsto efficiently manage keywords In spite of the implicationsand advantage of text mining [17] this study has limitationsFirst the research range is limited to two airlines in Korea

and Japan Secondly since the review data are producedspontaneously by clients collected data can be biased andlimited to active clients who are willing to express hisherexperience erefore in future study selecting more rep-resentative airlines is needed and also minimizing conve-nience of the respondents is needed to generalize the resultsof the study in the future

Data Availability

e text mining data used to support the findings of thisstudy are available from the authors upon request e dataused in this study are Airline Reviews and Rating data fromSkytrax (httpwwwairlinequalitycom)

Conflicts of Interest

e authors declare that they have no conflicts of interest

References

[1] P Nayak and A K Chatterjee ldquoEffects of aluminium exposureon brain glutamate and GABA systems an experimental studyin ratsrdquo Food and Chemical Toxicology vol 39 no 12pp 1285ndash1289 2001

[2] J A Chevalier and D Mayzlin ldquoe effect of word of mouthon sales online book reviewsrdquo Journal of Marketing Researchvol 43 no 3 pp 345ndash354 2018

[3] W Fan LWallace S Rich and Z Zhang ldquoTapping the powerof text miningrdquo Communications of the ACM vol 49 no 9pp 76ndash82 2006

[4] D Meyer K Hornik and I Feinerer ldquoText mining in-frastructure in Rrdquo Journal of Statistical Software vol 25 no 5pp 1ndash54 2008

[5] R V Kozinets K de Valck A CWojnicki and S J SWilnerldquoNetworked narratives understanding word-of-mouth mar-keting in online communitiesrdquo Journal of Marketing vol 74no 2 pp 71ndash89 2010

[6] A Hotho A Nurnberger and G Paaszlig ldquoA brief survey of textminingrdquo Ldv Forum vol 20 no 1 pp 19ndash62 2005

[7] O Netzer R Feldman J Goldenberg and M Fresko ldquoMineyour own business market-structure surveillance through textminingrdquo Marketing Science vol 31 no 3 pp 521ndash543 2012

[8] M M Mostafa ldquoMore than words social networksrsquo textmining for consumer brand sentimentsrdquo Expert Systems withApplications vol 40 no 10 pp 4241ndash4251 2013

[9] D Paranyushkin ldquoText network analysis (2010)rdquo in Con-ference du Performing Arts Forum 2011 httpnoduslabscomresearchpathways-meaning-circulation

[10] R Feldman and J Sanger gte Text Mining Handbook Ad-vanced Approaches in Analysing Un-Structured Data Cam-bridge University Press Cambridge UK 2007

[11] R W Schmenner ldquoHow can service businesses survive andprosperrdquo SloanManagement Review1986-1998 vol 27 no 3p 21 1986

[12] J Fitzsimmons and M Fitzsimmons Service managementOperations Strategy Information Technology McGraw-HillHigher Education New York City PA USA 2013

[13] S Jia ldquoBehind the ratings text mining of restaurant cus-tomersrsquo online reviewsrdquo International Journal of MarketResearch vol 60 no 6 pp 561ndash572 2018

[14] J-W Hong and S-B Park ldquoStudy on the extraction of corekeywords and its effects through text miningrdquo International

Mobile Information Systems 7

after the cluster was set In this method (1) generate k initialseed randomly within data domain (2) create k clusters byassociating every observation with the nearest seed positionand (3) adjust the center position of each of the k clustersusing the average of the observations belonging to thecluster By repeating this process until all observations areassociated it can make clusters which have similar obser-vations corresponding to the number of k

Step 1 (cluster setting) Set up a cluster with random seeds inthe data and the cluster is grouped by the Euclidean distancebetween each data and seed based on near distance

min S 1113944k

i11113944xisinsi

xminus ui

2

(3)

where x is the words set x1 x2 xn1113864 1113865 s is the clusters sets1 s2 sk1113864 1113865 and ui is the average of cluster si

Step 2 (cluster adjust) Reset the value of cluster and adjustthe position of cluster by using the average of the data in thecluster

4 Results

41 Keywords Extraction by Text Mining In this studyseveral preprocessing steps were performed to improve theaccuracy of text mining

First all uppercase letters in text data are converted tolowercase letters to unify words Also we removed un-necessary special characters such as ldquordquo ldquordquo and so on Andthe definite article (ldquotherdquo) the indefinite article (ldquoardquo ldquoanrdquo)prepositions (ldquoofrdquo ldquoinrdquo ldquoforrdquo ldquothroughrdquo etc) and pronouns(ldquoitrdquo ldquotheirrdquo ldquohisrdquo etc) were also removed rough thisprocess a total of 3774 keywords were extracted e key-words were rearranged based on the sparsity which representsthe ratio of the whitespace in the matrix because analyzing allkeywords makes it difficult to explain meaningful results

ere were 11 keywords based on the sparsity 08 19keywords based on the sparsity 085 45 keywords based onthe sparsity 09 and 132 keywords based on the sparsity 095In this study a total of 45 keywords were extracted byapplying the sparsity 09 and it is used for analysis Too few

keywords are insufficient to understand customer behaviourbut too many keywords can distort results with less im-portant words In this analysis 45 keywords were selectedconsidering the usefulness of the interpretation whilereflecting the customer behaviour

e keywords with high frequency were ldquogoodrdquo ldquoseatsrdquoldquocabinrdquo ldquoclassrdquo ldquoexcellentrdquo ldquocomfortablerdquo ldquostaffrdquo and soon Figure 2 shows the frequency distribution of thesekeywords and Figure 3 shows the frequency of keywordappearance by the word cloud method

42 Keywords Classification by Text Clustering In this studywe performed keyword classification analysis for semanticanalysis of extracted 45 keywords e results are as followsKeyword classification was based on hierarchical clusteringand the distance between the keywords was measured by theEuclidean method

As a result of clustering analysis 45 keywords wereclassified into two clusters Cluster 1 consisted of servicecontent such as ldquoseatsrdquo ldquocabinrdquo ldquoclassrdquo and ldquostaffrdquo whoprovide the service And it has service evaluation-relatedkeywords such as ldquogoodrdquo ldquoexcellentrdquo ldquocomfortablerdquo etc

On the other hand Cluster 2 consisted of more detailservice content such as ldquomealrdquo ldquomovierdquo ldquodrinksrdquo ldquocheckrdquoldquoservedrdquo and so on Also it has service evaluation-relatedkeywords such as ldquogreatrdquo ldquonicerdquo ldquowellrdquo ldquobestrdquo ldquocleanrdquo etc(see Figure 4)

Figure 5 shows the result of keyword visualization withtwo clusters based on the k-means method for better insighte keywords in Cluster 1 are formed to be spaced apartfrom each other and the keywords in Cluster 2 are formed tobe relatively wide spacing And the two componentsexplained 6183 of the point variability

43 Effect of Core Keywords In this study we performedcombining analysis with core keywords data extracted bytext mining and questionnaire respondent data to overcomethe disadvantages of existing text mining research and toprovide practical implications to the aviation industry Inother words since core keywords appear as a result of textmining and cluster analysis can be considered to representonline reviews of airline customers we attempted to un-derstand the meaning of representative keywords by ex-amining the influence of each evaluation concept oncustomer satisfaction and customer recommendation basedon the evaluation of these customers

In the analysis we analyzed the effect of the four key-words such as ldquoseatsrdquo ldquostaffrdquo ldquocabinrdquordquo and ldquoclassrdquo which arethe main keywords corresponding to the contents of theCluster 1 on customer satisfaction and recommendation ereason for this is that adjective and adverbial keywords amongthe keywords of Cluster 1 are mainly keywords indicating theresult of the service And the keywords appearing in Cluster 2are keywords that deal with details of the service contents ofCluster 1 So we define that these keywords are conceptuallyincluded in the above four top keywords

In the analytical model the questionnaire itemswere used as variables Independent variables were the

Table 1 Characteristics of country

No CountryKorea Japan Total

N N N 1 United States 58 294 81 379 139 3382 Australia 50 254 2 09 52 1273 United Kingdom 25 127 19 89 44 1074 Canada 4 20 21 98 25 615 Singapore 3 15 18 84 21 516 Japan 3 15 15 70 18 447 South Korea 13 66 1 05 14 348 Brazil 1 05 12 56 13 329 Germany 7 36 4 19 11 2710 New Zealand 8 41 1 05 9 22

4 Mobile Information Systems

questionnaire items for service evaluation on seats staffcabin and class And dependent variables were satisfac-tion and recommendation after using airline service

As a result of analysis seats staff and cabin were foundto be core keywords which had a significant effect on

satisfaction but class did not (see Table 2) In other wordsthe WOM (word of mouth) of the customers related to seatand staff services influence directly on customer satisfactionOn the other hand the customers of the airline talk fre-quently about class type such as first class and business classbut the class does not affect the customer satisfaction

is means that the airline should pay particular at-tention to the seat service and the staffrsquos efforts are needed tocomplement their service It also suggests that customersatisfaction can be improved by paying attention to theservice environment factors in the cabin

On the other hand seats staff cabin and class have asignificant effect on recommendation And in terms of powerof influence seats service was found to have a greater effect onrecommendation than staff serviceese results indicate thatseat service is also an important factor in customer recom-mendation and the service of airline crew can complement itHowever cabin and class service did not have a significanteffect on recommendation (Table 3) ese results indicatethat although airline passengers talk a lot about the services inthe cabin they do not really have much to do with customerreferrals on recommendation As a result it is suggested thatseats and staff services are especially required to provideairline with recommendations from their customers

5 Conclusion

Data mining techniques need to be used in the aviationindustry to understand consumer behaviour at is this

0

100

200

Word

Freq

Always

Atte

ntive

Best

Bette

rBit

Boarding

Busin

ess

Cabin

Check

Checkin

Choice

Class

Clean

Comfortable

Definitely

Drin

ksEcon

omy

Efficient

Entertainm

ent

Excellent

Experie

nce

Flat

Friend

lyFu

llGoo

dGreat

Groun

dHelpful

How

ever

Infligh

tLeg

Legs

Like

Limited

Loun

geMade

Many

Meal

Meals

Movies

Much

Nice

Offe

red

Onb

oard

Overall

Polite

Quality

Quite

Really

Screen

Seats

Selection

Served

Sleep

Space

Staff

Well

Figure 2 Distribution of keyword frequency

Figure 3 Word cloud of keywords

Mobile Information Systems 5

Great

Meal

Loun

geNice

Entertainm

ent

Infligh

tWell

Best

Experie

nce

Really

Meals

Served

Friend

lyHow

ever leg

Sleep

Overall

Quite bit

Bette

rMuch

Quality

Movies

Selection

Check

Made

Offered

Boarding

Clean

Onb

oard

Efficient

Helpful

Polite

Atte

ntive

Choice

Drin

ksGoo

dSeats

Staff

Comfortable

Excellent

Cabin

Econ

omy

Busin

ess

Class

5

10

15

20

25

30Heigh

t

Figure 4 Cluster dendrogram

ndash15 ndash10 ndash5 0 5

ndash4

ndash2

0

2

4

6

Component 1

Com

pone

nt 2

AttentiveBest Better

BitBoarding

Business

CabinCheck Choice

Class

CleanComfortable

DrinksEconomy

Efficient

EntertainmentExcellent

Experience

Friendly

Good

Great

HelpfulHoweverInflightLeg

Lounge

MadeMeal

Meals

Movies

Much

Nice

OfferedOnboardOverallPolite

Quality

QuiteReally

Seats SelectionServed Sleep

Staff

Well

1

2

Figure 5 Cluster plot ese two components explain 6183 of the point variability

Table 2 Effects of core keyword on satisfaction

Variables B β t-value p-valueSeats 0987 0513 186lowastlowastlowast 0001Staff 0874 0334 112lowastlowastlowast 0001Cabin 0555 0235 74lowastlowastlowast 0001Class 0056 0011 05 0645lowastlowastlowastρlt 0001 F value 3424 R2 0771

Table 3 Effects of core keyword on recommendation

Variables B β t-value p-valueSeats 0156 0536 135lowastlowastlowast 0001Staff 0110 0275 64lowastlowastlowast 0001Cabin 0020 0055 12 0224Class 0039 0051 15 0139lowastlowastlowastρlt 0001 F value 1110 R2 0518

6 Mobile Information Systems

study used text mining to develop a unified understanding ofkeywords in the aviation industry in a data-driven wayBased on the airline review data we proposed a two-stepprocess of extracting key keywords by text mining andgrouping them into cluster analysis Specifically we useda combination of metrics and clustering algorithms topreprocess and analyze text data related to keywords ex-traction method including text from the scientific literatureand news articles is study seeks firstly for identifyingprominent keywords at consumersrsquo side using text miningmethod on consumersrsquo online review data and then forconfirming influences that the keywords affect corporatemarketing performance is study is not only the researchthat searches for key keywords [16] but also the researchthat identifies marketing performance in the aviation in-dustry through text mining in service category level Con-clusion and implication are as follows

First keywords are shown to have distinctive clustercharacteristics As a result of identifying characteristics ofmajor keywords through clustering the keywords have beenclassified into three sections including service that airlinesprovides with details of each service and assessment on theservices In other words since the service provided by airlineand the details of each service are differentiated at differentlevels it indicates that service management should becentered on core service elements that can be clearly rec-ognized in order to improve service evaluation

Second the key keywords extracted from text miningwere found to have a different relationship with corporateperformance In other words the service of the seats and thestaff was more important to the company performance thanthe cabin or class ese results show that comfort is the keycustomerrsquos needs in long-distance air travel and that theservice focused on the comfort of the seat or the comfort ofthe seat is an important factor in corporate performance

ird recommendation and satisfaction should bemanaged distinctively in the service management of theaviation industry e results show that keywords that affectconsumersrsquo satisfaction and keywords that affect consumersrsquorecommendation are found to differ and the degree ofimpacts is also shown to be different as well In other wordsseats staff and cabin are important factors to improve thesatisfaction and recommendation of consumers Seats arethe most important factors in consumer satisfaction andrecommendation but relatively more staff and cabin areimportant for consumer satisfaction and seats are moreimportant for consumer recommendation ese resultsshow that the human service of the crew which can berelatively subjective is more influential in satisfaction Onthe other hand seats are more important for relativelyobjective recommendations

is study presents academic implication that the studyhas extended its application area of text mining It hascurrently focused on exploratory study while this study hasextended it to study field of cause and effect Moreover theresearch also presents practical implication for corporationsto efficiently manage keywords In spite of the implicationsand advantage of text mining [17] this study has limitationsFirst the research range is limited to two airlines in Korea

and Japan Secondly since the review data are producedspontaneously by clients collected data can be biased andlimited to active clients who are willing to express hisherexperience erefore in future study selecting more rep-resentative airlines is needed and also minimizing conve-nience of the respondents is needed to generalize the resultsof the study in the future

Data Availability

e text mining data used to support the findings of thisstudy are available from the authors upon request e dataused in this study are Airline Reviews and Rating data fromSkytrax (httpwwwairlinequalitycom)

Conflicts of Interest

e authors declare that they have no conflicts of interest

References

[1] P Nayak and A K Chatterjee ldquoEffects of aluminium exposureon brain glutamate and GABA systems an experimental studyin ratsrdquo Food and Chemical Toxicology vol 39 no 12pp 1285ndash1289 2001

[2] J A Chevalier and D Mayzlin ldquoe effect of word of mouthon sales online book reviewsrdquo Journal of Marketing Researchvol 43 no 3 pp 345ndash354 2018

[3] W Fan LWallace S Rich and Z Zhang ldquoTapping the powerof text miningrdquo Communications of the ACM vol 49 no 9pp 76ndash82 2006

[4] D Meyer K Hornik and I Feinerer ldquoText mining in-frastructure in Rrdquo Journal of Statistical Software vol 25 no 5pp 1ndash54 2008

[5] R V Kozinets K de Valck A CWojnicki and S J SWilnerldquoNetworked narratives understanding word-of-mouth mar-keting in online communitiesrdquo Journal of Marketing vol 74no 2 pp 71ndash89 2010

[6] A Hotho A Nurnberger and G Paaszlig ldquoA brief survey of textminingrdquo Ldv Forum vol 20 no 1 pp 19ndash62 2005

[7] O Netzer R Feldman J Goldenberg and M Fresko ldquoMineyour own business market-structure surveillance through textminingrdquo Marketing Science vol 31 no 3 pp 521ndash543 2012

[8] M M Mostafa ldquoMore than words social networksrsquo textmining for consumer brand sentimentsrdquo Expert Systems withApplications vol 40 no 10 pp 4241ndash4251 2013

[9] D Paranyushkin ldquoText network analysis (2010)rdquo in Con-ference du Performing Arts Forum 2011 httpnoduslabscomresearchpathways-meaning-circulation

[10] R Feldman and J Sanger gte Text Mining Handbook Ad-vanced Approaches in Analysing Un-Structured Data Cam-bridge University Press Cambridge UK 2007

[11] R W Schmenner ldquoHow can service businesses survive andprosperrdquo SloanManagement Review1986-1998 vol 27 no 3p 21 1986

[12] J Fitzsimmons and M Fitzsimmons Service managementOperations Strategy Information Technology McGraw-HillHigher Education New York City PA USA 2013

[13] S Jia ldquoBehind the ratings text mining of restaurant cus-tomersrsquo online reviewsrdquo International Journal of MarketResearch vol 60 no 6 pp 561ndash572 2018

[14] J-W Hong and S-B Park ldquoStudy on the extraction of corekeywords and its effects through text miningrdquo International

Mobile Information Systems 7

questionnaire items for service evaluation on seats staffcabin and class And dependent variables were satisfac-tion and recommendation after using airline service

As a result of analysis seats staff and cabin were foundto be core keywords which had a significant effect on

satisfaction but class did not (see Table 2) In other wordsthe WOM (word of mouth) of the customers related to seatand staff services influence directly on customer satisfactionOn the other hand the customers of the airline talk fre-quently about class type such as first class and business classbut the class does not affect the customer satisfaction

is means that the airline should pay particular at-tention to the seat service and the staffrsquos efforts are needed tocomplement their service It also suggests that customersatisfaction can be improved by paying attention to theservice environment factors in the cabin

On the other hand seats staff cabin and class have asignificant effect on recommendation And in terms of powerof influence seats service was found to have a greater effect onrecommendation than staff serviceese results indicate thatseat service is also an important factor in customer recom-mendation and the service of airline crew can complement itHowever cabin and class service did not have a significanteffect on recommendation (Table 3) ese results indicatethat although airline passengers talk a lot about the services inthe cabin they do not really have much to do with customerreferrals on recommendation As a result it is suggested thatseats and staff services are especially required to provideairline with recommendations from their customers

5 Conclusion

Data mining techniques need to be used in the aviationindustry to understand consumer behaviour at is this

0

100

200

Word

Freq

Always

Atte

ntive

Best

Bette

rBit

Boarding

Busin

ess

Cabin

Check

Checkin

Choice

Class

Clean

Comfortable

Definitely

Drin

ksEcon

omy

Efficient

Entertainm

ent

Excellent

Experie

nce

Flat

Friend

lyFu

llGoo

dGreat

Groun

dHelpful

How

ever

Infligh

tLeg

Legs

Like

Limited

Loun

geMade

Many

Meal

Meals

Movies

Much

Nice

Offe

red

Onb

oard

Overall

Polite

Quality

Quite

Really

Screen

Seats

Selection

Served

Sleep

Space

Staff

Well

Figure 2 Distribution of keyword frequency

Figure 3 Word cloud of keywords

Mobile Information Systems 5

Great

Meal

Loun

geNice

Entertainm

ent

Infligh

tWell

Best

Experie

nce

Really

Meals

Served

Friend

lyHow

ever leg

Sleep

Overall

Quite bit

Bette

rMuch

Quality

Movies

Selection

Check

Made

Offered

Boarding

Clean

Onb

oard

Efficient

Helpful

Polite

Atte

ntive

Choice

Drin

ksGoo

dSeats

Staff

Comfortable

Excellent

Cabin

Econ

omy

Busin

ess

Class

5

10

15

20

25

30Heigh

t

Figure 4 Cluster dendrogram

ndash15 ndash10 ndash5 0 5

ndash4

ndash2

0

2

4

6

Component 1

Com

pone

nt 2

AttentiveBest Better

BitBoarding

Business

CabinCheck Choice

Class

CleanComfortable

DrinksEconomy

Efficient

EntertainmentExcellent

Experience

Friendly

Good

Great

HelpfulHoweverInflightLeg

Lounge

MadeMeal

Meals

Movies

Much

Nice

OfferedOnboardOverallPolite

Quality

QuiteReally

Seats SelectionServed Sleep

Staff

Well

1

2

Figure 5 Cluster plot ese two components explain 6183 of the point variability

Table 2 Effects of core keyword on satisfaction

Variables B β t-value p-valueSeats 0987 0513 186lowastlowastlowast 0001Staff 0874 0334 112lowastlowastlowast 0001Cabin 0555 0235 74lowastlowastlowast 0001Class 0056 0011 05 0645lowastlowastlowastρlt 0001 F value 3424 R2 0771

Table 3 Effects of core keyword on recommendation

Variables B β t-value p-valueSeats 0156 0536 135lowastlowastlowast 0001Staff 0110 0275 64lowastlowastlowast 0001Cabin 0020 0055 12 0224Class 0039 0051 15 0139lowastlowastlowastρlt 0001 F value 1110 R2 0518

6 Mobile Information Systems

study used text mining to develop a unified understanding ofkeywords in the aviation industry in a data-driven wayBased on the airline review data we proposed a two-stepprocess of extracting key keywords by text mining andgrouping them into cluster analysis Specifically we useda combination of metrics and clustering algorithms topreprocess and analyze text data related to keywords ex-traction method including text from the scientific literatureand news articles is study seeks firstly for identifyingprominent keywords at consumersrsquo side using text miningmethod on consumersrsquo online review data and then forconfirming influences that the keywords affect corporatemarketing performance is study is not only the researchthat searches for key keywords [16] but also the researchthat identifies marketing performance in the aviation in-dustry through text mining in service category level Con-clusion and implication are as follows

First keywords are shown to have distinctive clustercharacteristics As a result of identifying characteristics ofmajor keywords through clustering the keywords have beenclassified into three sections including service that airlinesprovides with details of each service and assessment on theservices In other words since the service provided by airlineand the details of each service are differentiated at differentlevels it indicates that service management should becentered on core service elements that can be clearly rec-ognized in order to improve service evaluation

Second the key keywords extracted from text miningwere found to have a different relationship with corporateperformance In other words the service of the seats and thestaff was more important to the company performance thanthe cabin or class ese results show that comfort is the keycustomerrsquos needs in long-distance air travel and that theservice focused on the comfort of the seat or the comfort ofthe seat is an important factor in corporate performance

ird recommendation and satisfaction should bemanaged distinctively in the service management of theaviation industry e results show that keywords that affectconsumersrsquo satisfaction and keywords that affect consumersrsquorecommendation are found to differ and the degree ofimpacts is also shown to be different as well In other wordsseats staff and cabin are important factors to improve thesatisfaction and recommendation of consumers Seats arethe most important factors in consumer satisfaction andrecommendation but relatively more staff and cabin areimportant for consumer satisfaction and seats are moreimportant for consumer recommendation ese resultsshow that the human service of the crew which can berelatively subjective is more influential in satisfaction Onthe other hand seats are more important for relativelyobjective recommendations

is study presents academic implication that the studyhas extended its application area of text mining It hascurrently focused on exploratory study while this study hasextended it to study field of cause and effect Moreover theresearch also presents practical implication for corporationsto efficiently manage keywords In spite of the implicationsand advantage of text mining [17] this study has limitationsFirst the research range is limited to two airlines in Korea

and Japan Secondly since the review data are producedspontaneously by clients collected data can be biased andlimited to active clients who are willing to express hisherexperience erefore in future study selecting more rep-resentative airlines is needed and also minimizing conve-nience of the respondents is needed to generalize the resultsof the study in the future

Data Availability

e text mining data used to support the findings of thisstudy are available from the authors upon request e dataused in this study are Airline Reviews and Rating data fromSkytrax (httpwwwairlinequalitycom)

Conflicts of Interest

e authors declare that they have no conflicts of interest

References

[1] P Nayak and A K Chatterjee ldquoEffects of aluminium exposureon brain glutamate and GABA systems an experimental studyin ratsrdquo Food and Chemical Toxicology vol 39 no 12pp 1285ndash1289 2001

[2] J A Chevalier and D Mayzlin ldquoe effect of word of mouthon sales online book reviewsrdquo Journal of Marketing Researchvol 43 no 3 pp 345ndash354 2018

[3] W Fan LWallace S Rich and Z Zhang ldquoTapping the powerof text miningrdquo Communications of the ACM vol 49 no 9pp 76ndash82 2006

[4] D Meyer K Hornik and I Feinerer ldquoText mining in-frastructure in Rrdquo Journal of Statistical Software vol 25 no 5pp 1ndash54 2008

[5] R V Kozinets K de Valck A CWojnicki and S J SWilnerldquoNetworked narratives understanding word-of-mouth mar-keting in online communitiesrdquo Journal of Marketing vol 74no 2 pp 71ndash89 2010

[6] A Hotho A Nurnberger and G Paaszlig ldquoA brief survey of textminingrdquo Ldv Forum vol 20 no 1 pp 19ndash62 2005

[7] O Netzer R Feldman J Goldenberg and M Fresko ldquoMineyour own business market-structure surveillance through textminingrdquo Marketing Science vol 31 no 3 pp 521ndash543 2012

[8] M M Mostafa ldquoMore than words social networksrsquo textmining for consumer brand sentimentsrdquo Expert Systems withApplications vol 40 no 10 pp 4241ndash4251 2013

[9] D Paranyushkin ldquoText network analysis (2010)rdquo in Con-ference du Performing Arts Forum 2011 httpnoduslabscomresearchpathways-meaning-circulation

[10] R Feldman and J Sanger gte Text Mining Handbook Ad-vanced Approaches in Analysing Un-Structured Data Cam-bridge University Press Cambridge UK 2007

[11] R W Schmenner ldquoHow can service businesses survive andprosperrdquo SloanManagement Review1986-1998 vol 27 no 3p 21 1986

[12] J Fitzsimmons and M Fitzsimmons Service managementOperations Strategy Information Technology McGraw-HillHigher Education New York City PA USA 2013

[13] S Jia ldquoBehind the ratings text mining of restaurant cus-tomersrsquo online reviewsrdquo International Journal of MarketResearch vol 60 no 6 pp 561ndash572 2018

[14] J-W Hong and S-B Park ldquoStudy on the extraction of corekeywords and its effects through text miningrdquo International

Mobile Information Systems 7

Great

Meal

Loun

geNice

Entertainm

ent

Infligh

tWell

Best

Experie

nce

Really

Meals

Served

Friend

lyHow

ever leg

Sleep

Overall

Quite bit

Bette

rMuch

Quality

Movies

Selection

Check

Made

Offered

Boarding

Clean

Onb

oard

Efficient

Helpful

Polite

Atte

ntive

Choice

Drin

ksGoo

dSeats

Staff

Comfortable

Excellent

Cabin

Econ

omy

Busin

ess

Class

5

10

15

20

25

30Heigh

t

Figure 4 Cluster dendrogram

ndash15 ndash10 ndash5 0 5

ndash4

ndash2

0

2

4

6

Component 1

Com

pone

nt 2

AttentiveBest Better

BitBoarding

Business

CabinCheck Choice

Class

CleanComfortable

DrinksEconomy

Efficient

EntertainmentExcellent

Experience

Friendly

Good

Great

HelpfulHoweverInflightLeg

Lounge

MadeMeal

Meals

Movies

Much

Nice

OfferedOnboardOverallPolite

Quality

QuiteReally

Seats SelectionServed Sleep

Staff

Well

1

2

Figure 5 Cluster plot ese two components explain 6183 of the point variability

Table 2 Effects of core keyword on satisfaction

Variables B β t-value p-valueSeats 0987 0513 186lowastlowastlowast 0001Staff 0874 0334 112lowastlowastlowast 0001Cabin 0555 0235 74lowastlowastlowast 0001Class 0056 0011 05 0645lowastlowastlowastρlt 0001 F value 3424 R2 0771

Table 3 Effects of core keyword on recommendation

Variables B β t-value p-valueSeats 0156 0536 135lowastlowastlowast 0001Staff 0110 0275 64lowastlowastlowast 0001Cabin 0020 0055 12 0224Class 0039 0051 15 0139lowastlowastlowastρlt 0001 F value 1110 R2 0518

6 Mobile Information Systems

study used text mining to develop a unified understanding ofkeywords in the aviation industry in a data-driven wayBased on the airline review data we proposed a two-stepprocess of extracting key keywords by text mining andgrouping them into cluster analysis Specifically we useda combination of metrics and clustering algorithms topreprocess and analyze text data related to keywords ex-traction method including text from the scientific literatureand news articles is study seeks firstly for identifyingprominent keywords at consumersrsquo side using text miningmethod on consumersrsquo online review data and then forconfirming influences that the keywords affect corporatemarketing performance is study is not only the researchthat searches for key keywords [16] but also the researchthat identifies marketing performance in the aviation in-dustry through text mining in service category level Con-clusion and implication are as follows

First keywords are shown to have distinctive clustercharacteristics As a result of identifying characteristics ofmajor keywords through clustering the keywords have beenclassified into three sections including service that airlinesprovides with details of each service and assessment on theservices In other words since the service provided by airlineand the details of each service are differentiated at differentlevels it indicates that service management should becentered on core service elements that can be clearly rec-ognized in order to improve service evaluation

Second the key keywords extracted from text miningwere found to have a different relationship with corporateperformance In other words the service of the seats and thestaff was more important to the company performance thanthe cabin or class ese results show that comfort is the keycustomerrsquos needs in long-distance air travel and that theservice focused on the comfort of the seat or the comfort ofthe seat is an important factor in corporate performance

ird recommendation and satisfaction should bemanaged distinctively in the service management of theaviation industry e results show that keywords that affectconsumersrsquo satisfaction and keywords that affect consumersrsquorecommendation are found to differ and the degree ofimpacts is also shown to be different as well In other wordsseats staff and cabin are important factors to improve thesatisfaction and recommendation of consumers Seats arethe most important factors in consumer satisfaction andrecommendation but relatively more staff and cabin areimportant for consumer satisfaction and seats are moreimportant for consumer recommendation ese resultsshow that the human service of the crew which can berelatively subjective is more influential in satisfaction Onthe other hand seats are more important for relativelyobjective recommendations

is study presents academic implication that the studyhas extended its application area of text mining It hascurrently focused on exploratory study while this study hasextended it to study field of cause and effect Moreover theresearch also presents practical implication for corporationsto efficiently manage keywords In spite of the implicationsand advantage of text mining [17] this study has limitationsFirst the research range is limited to two airlines in Korea

and Japan Secondly since the review data are producedspontaneously by clients collected data can be biased andlimited to active clients who are willing to express hisherexperience erefore in future study selecting more rep-resentative airlines is needed and also minimizing conve-nience of the respondents is needed to generalize the resultsof the study in the future

Data Availability

e text mining data used to support the findings of thisstudy are available from the authors upon request e dataused in this study are Airline Reviews and Rating data fromSkytrax (httpwwwairlinequalitycom)

Conflicts of Interest

e authors declare that they have no conflicts of interest

References

[1] P Nayak and A K Chatterjee ldquoEffects of aluminium exposureon brain glutamate and GABA systems an experimental studyin ratsrdquo Food and Chemical Toxicology vol 39 no 12pp 1285ndash1289 2001

[2] J A Chevalier and D Mayzlin ldquoe effect of word of mouthon sales online book reviewsrdquo Journal of Marketing Researchvol 43 no 3 pp 345ndash354 2018

[3] W Fan LWallace S Rich and Z Zhang ldquoTapping the powerof text miningrdquo Communications of the ACM vol 49 no 9pp 76ndash82 2006

[4] D Meyer K Hornik and I Feinerer ldquoText mining in-frastructure in Rrdquo Journal of Statistical Software vol 25 no 5pp 1ndash54 2008

[5] R V Kozinets K de Valck A CWojnicki and S J SWilnerldquoNetworked narratives understanding word-of-mouth mar-keting in online communitiesrdquo Journal of Marketing vol 74no 2 pp 71ndash89 2010

[6] A Hotho A Nurnberger and G Paaszlig ldquoA brief survey of textminingrdquo Ldv Forum vol 20 no 1 pp 19ndash62 2005

[7] O Netzer R Feldman J Goldenberg and M Fresko ldquoMineyour own business market-structure surveillance through textminingrdquo Marketing Science vol 31 no 3 pp 521ndash543 2012

[8] M M Mostafa ldquoMore than words social networksrsquo textmining for consumer brand sentimentsrdquo Expert Systems withApplications vol 40 no 10 pp 4241ndash4251 2013

[9] D Paranyushkin ldquoText network analysis (2010)rdquo in Con-ference du Performing Arts Forum 2011 httpnoduslabscomresearchpathways-meaning-circulation

[10] R Feldman and J Sanger gte Text Mining Handbook Ad-vanced Approaches in Analysing Un-Structured Data Cam-bridge University Press Cambridge UK 2007

[11] R W Schmenner ldquoHow can service businesses survive andprosperrdquo SloanManagement Review1986-1998 vol 27 no 3p 21 1986

[12] J Fitzsimmons and M Fitzsimmons Service managementOperations Strategy Information Technology McGraw-HillHigher Education New York City PA USA 2013

[13] S Jia ldquoBehind the ratings text mining of restaurant cus-tomersrsquo online reviewsrdquo International Journal of MarketResearch vol 60 no 6 pp 561ndash572 2018

[14] J-W Hong and S-B Park ldquoStudy on the extraction of corekeywords and its effects through text miningrdquo International

Mobile Information Systems 7

study used text mining to develop a unified understanding ofkeywords in the aviation industry in a data-driven wayBased on the airline review data we proposed a two-stepprocess of extracting key keywords by text mining andgrouping them into cluster analysis Specifically we useda combination of metrics and clustering algorithms topreprocess and analyze text data related to keywords ex-traction method including text from the scientific literatureand news articles is study seeks firstly for identifyingprominent keywords at consumersrsquo side using text miningmethod on consumersrsquo online review data and then forconfirming influences that the keywords affect corporatemarketing performance is study is not only the researchthat searches for key keywords [16] but also the researchthat identifies marketing performance in the aviation in-dustry through text mining in service category level Con-clusion and implication are as follows

First keywords are shown to have distinctive clustercharacteristics As a result of identifying characteristics ofmajor keywords through clustering the keywords have beenclassified into three sections including service that airlinesprovides with details of each service and assessment on theservices In other words since the service provided by airlineand the details of each service are differentiated at differentlevels it indicates that service management should becentered on core service elements that can be clearly rec-ognized in order to improve service evaluation

Second the key keywords extracted from text miningwere found to have a different relationship with corporateperformance In other words the service of the seats and thestaff was more important to the company performance thanthe cabin or class ese results show that comfort is the keycustomerrsquos needs in long-distance air travel and that theservice focused on the comfort of the seat or the comfort ofthe seat is an important factor in corporate performance

ird recommendation and satisfaction should bemanaged distinctively in the service management of theaviation industry e results show that keywords that affectconsumersrsquo satisfaction and keywords that affect consumersrsquorecommendation are found to differ and the degree ofimpacts is also shown to be different as well In other wordsseats staff and cabin are important factors to improve thesatisfaction and recommendation of consumers Seats arethe most important factors in consumer satisfaction andrecommendation but relatively more staff and cabin areimportant for consumer satisfaction and seats are moreimportant for consumer recommendation ese resultsshow that the human service of the crew which can berelatively subjective is more influential in satisfaction Onthe other hand seats are more important for relativelyobjective recommendations

is study presents academic implication that the studyhas extended its application area of text mining It hascurrently focused on exploratory study while this study hasextended it to study field of cause and effect Moreover theresearch also presents practical implication for corporationsto efficiently manage keywords In spite of the implicationsand advantage of text mining [17] this study has limitationsFirst the research range is limited to two airlines in Korea

and Japan Secondly since the review data are producedspontaneously by clients collected data can be biased andlimited to active clients who are willing to express hisherexperience erefore in future study selecting more rep-resentative airlines is needed and also minimizing conve-nience of the respondents is needed to generalize the resultsof the study in the future

Data Availability

e text mining data used to support the findings of thisstudy are available from the authors upon request e dataused in this study are Airline Reviews and Rating data fromSkytrax (httpwwwairlinequalitycom)

Conflicts of Interest

e authors declare that they have no conflicts of interest

References

[1] P Nayak and A K Chatterjee ldquoEffects of aluminium exposureon brain glutamate and GABA systems an experimental studyin ratsrdquo Food and Chemical Toxicology vol 39 no 12pp 1285ndash1289 2001

[2] J A Chevalier and D Mayzlin ldquoe effect of word of mouthon sales online book reviewsrdquo Journal of Marketing Researchvol 43 no 3 pp 345ndash354 2018

[3] W Fan LWallace S Rich and Z Zhang ldquoTapping the powerof text miningrdquo Communications of the ACM vol 49 no 9pp 76ndash82 2006

[4] D Meyer K Hornik and I Feinerer ldquoText mining in-frastructure in Rrdquo Journal of Statistical Software vol 25 no 5pp 1ndash54 2008

[5] R V Kozinets K de Valck A CWojnicki and S J SWilnerldquoNetworked narratives understanding word-of-mouth mar-keting in online communitiesrdquo Journal of Marketing vol 74no 2 pp 71ndash89 2010

[6] A Hotho A Nurnberger and G Paaszlig ldquoA brief survey of textminingrdquo Ldv Forum vol 20 no 1 pp 19ndash62 2005

[7] O Netzer R Feldman J Goldenberg and M Fresko ldquoMineyour own business market-structure surveillance through textminingrdquo Marketing Science vol 31 no 3 pp 521ndash543 2012

[8] M M Mostafa ldquoMore than words social networksrsquo textmining for consumer brand sentimentsrdquo Expert Systems withApplications vol 40 no 10 pp 4241ndash4251 2013

[9] D Paranyushkin ldquoText network analysis (2010)rdquo in Con-ference du Performing Arts Forum 2011 httpnoduslabscomresearchpathways-meaning-circulation

[10] R Feldman and J Sanger gte Text Mining Handbook Ad-vanced Approaches in Analysing Un-Structured Data Cam-bridge University Press Cambridge UK 2007

[11] R W Schmenner ldquoHow can service businesses survive andprosperrdquo SloanManagement Review1986-1998 vol 27 no 3p 21 1986

[12] J Fitzsimmons and M Fitzsimmons Service managementOperations Strategy Information Technology McGraw-HillHigher Education New York City PA USA 2013

[13] S Jia ldquoBehind the ratings text mining of restaurant cus-tomersrsquo online reviewsrdquo International Journal of MarketResearch vol 60 no 6 pp 561ndash572 2018

[14] J-W Hong and S-B Park ldquoStudy on the extraction of corekeywords and its effects through text miningrdquo International

Mobile Information Systems 7

Journal of Web Science and Engineeringfor Smart Devicesvol 3 no 2 pp 7ndash12 2016

[15] D Jung and S B Kim ldquoe effects of information quality ofmobile on consumer behaviour intension for airlinerdquo Asia-pacific Journal of Multimedia Services Convergent with ArtHumanities and Sociology vol 6 no 9 pp 19ndash26 2016

[16] S-B Park and J-W Hong ldquoRelationship of core keywordsand marketing performance obtained by using text mining acomparative studyrdquo New Physics Sae Mulli vol 67 no 5pp 562ndash568 2017

[17] X Wang and S Liu ldquoAnalysis and research of enterprisetechnology competent advantage on text mining and corre-spondence analysisrdquo International Journal of Databasegteoryand Application vol 6 no 5 pp 133ndash140 2013

8 Mobile Information Systems

Computer Games Technology

International Journal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom

Journal ofEngineeringVolume 2018

Advances in

FuzzySystems

Hindawiwwwhindawicom

Volume 2018

International Journal of

ReconfigurableComputing

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Applied Computational Intelligence and Soft Computing

thinspAdvancesthinspinthinsp

thinspArtificial Intelligence

Hindawiwwwhindawicom Volumethinsp2018

Hindawiwwwhindawicom Volume 2018

Civil EngineeringAdvances in

Hindawiwwwhindawicom Volume 2018

Electrical and Computer Engineering

Journal of

Journal of

Computer Networks and Communications

Hindawiwwwhindawicom Volume 2018

Hindawi

wwwhindawicom Volume 2018

Advances in

Multimedia

International Journal of

Biomedical Imaging

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Engineering Mathematics

International Journal of

RoboticsJournal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Computational Intelligence and Neuroscience

Hindawiwwwhindawicom Volume 2018

Mathematical Problems in Engineering

Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018

Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom

The Scientific World Journal

Volume 2018

Hindawiwwwhindawicom Volume 2018

Human-ComputerInteraction

Advances in

Hindawiwwwhindawicom Volume 2018

Scientic Programming

Submit your manuscripts atwwwhindawicom

Computer Games Technology

International Journal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom

Journal ofEngineeringVolume 2018

Advances in

FuzzySystems

Hindawiwwwhindawicom

Volume 2018

International Journal of

ReconfigurableComputing

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Applied Computational Intelligence and Soft Computing

thinspAdvancesthinspinthinsp

thinspArtificial Intelligence

Hindawiwwwhindawicom Volumethinsp2018

Hindawiwwwhindawicom Volume 2018

Civil EngineeringAdvances in

Hindawiwwwhindawicom Volume 2018

Electrical and Computer Engineering

Journal of

Journal of

Computer Networks and Communications

Hindawiwwwhindawicom Volume 2018

Hindawi

wwwhindawicom Volume 2018

Advances in

Multimedia

International Journal of

Biomedical Imaging

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Engineering Mathematics

International Journal of

RoboticsJournal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom Volume 2018

Computational Intelligence and Neuroscience

Hindawiwwwhindawicom Volume 2018

Mathematical Problems in Engineering

Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018

Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom

The Scientific World Journal

Volume 2018

Hindawiwwwhindawicom Volume 2018

Human-ComputerInteraction

Advances in

Hindawiwwwhindawicom Volume 2018

Scientic Programming

Submit your manuscripts atwwwhindawicom