21
Hybrid Mobile Apps in the Google Play Store: an Exploratory Investigation I.Malavolta, S.Ruberto, T.Soru, V.Terragni GSSI June 1, 2015 I.Malavolta, S.Ruberto, T.Soru, V.Terragni Hybrid Mobile Apps June 1, 2015 1 / 21

Presentazione

Embed Size (px)

Citation preview

Hybrid Mobile Apps in the Google Play Store:an Exploratory Investigation

I.Malavolta, S.Ruberto, T.Soru, V.Terragni

GSSI

June 1, 2015

I.Malavolta, S.Ruberto, T.Soru, V.Terragni Hybrid Mobile Apps June 1, 2015 1 / 21

Presentation structure

1 Context: Hybrid Mobile AppsTechnologiesIssues

2 Study DesignGoalsResearch questionsObjects and VariablesData Extraction

3 Findings4 Conclusion and Future works

Perceived ValuesConclusion

I.Malavolta, S.Ruberto, T.Soru, V.Terragni Hybrid Mobile Apps June 1, 2015 2 / 21

Context: Hybrid Mobile Apps Technologies

Hybrid Mobile AppsDevelopments technologies

• a possible solution to mobile platformsfragmentation

• consistent across platforms

• develop using standard web technologies:HTML5 CSS JAVASCRIPT

• on top of hybrid development frameworkI providing a native wrapperI a generic JavaScript API that bridges all

the service requests to the correspondingplatform API

I.Malavolta, S.Ruberto, T.Soru, V.Terragni Hybrid Mobile Apps June 1, 2015 3 / 21

Context: Hybrid Mobile Apps Issues

Pros and Cons

Pros

• cross-platform portability

• reuse of existing knowledge ofweb developers

• simpler and less expensivedevelopment processes

Cons

• restricted access to hardwarefeatures

• variations on user experience

• decrease in performance

Strong debate about benefits and drawbacks

Observation

As of today, limited empirical investigations have been performed onhybrid mobile apps

I.Malavolta, S.Ruberto, T.Soru, V.Terragni Hybrid Mobile Apps June 1, 2015 4 / 21

Study Design Goals

Research Goals

To analyse hybrid mobile apps

• in their actual context of use

• from both the developers’ and end users’viewpoints

• with a reproducible empirical strategyI well-defined empirical protocolI dataset comprising 11,917 apps and 3,041,315

user ratingsI dedicated analysis tool (available online)

1

1https://github.com/GabMar/ApkCategoryChecker

I.Malavolta, S.Ruberto, T.Soru, V.Terragni Hybrid Mobile Apps June 1, 2015 5 / 21

Study Design Research questions

Research questions

RQ1 Are hybrid mobile apps distributed in the Google Play Store?

RQ2 What are the most used hybrid development frameworks fordeveloping hybrid mobile apps?

RQ3 What are the most used 3rd-party web libraries for developing hybridmobile apps?

RQ4 How are hybrid mobile apps integrated to the Android platformand other installed apps?

RQ5 What is the di↵erence in the user perceived value between hybridand native mobile apps?

I.Malavolta, S.Ruberto, T.Soru, V.Terragni Hybrid Mobile Apps June 1, 2015 6 / 21

Study Design Objects and Variables

Objects and VariablesObjects

11,917 free Android apps from the Google Play Store

• large number of available apps

• easy to extract relevant information from app binaries

Reference set of hybrid development frameworkse.g., Apache Cordova, Appcelerator Titanium, etc.

Publicly available reference list on Wikipedia2 in order to

• have a starting point as complete and up to date as possible

• make our study replicable

2en.wikipedia.org/wiki/Multiplephoneweb-based_application_framework

I.Malavolta, S.Ruberto, T.Soru, V.Terragni Hybrid Mobile Apps June 1, 2015 7 / 21

Study Design Objects and Variables

Objects and VariablesVariables (1)

type identify whether the mobile app ishybrid or native.(RQ1)

frameworkName used hybrid development framework(e.g. Apache Cordova) (RQ2)

libraries identify the used specific 3rd-party weblibraries (RQ3)

I.Malavolta, S.Ruberto, T.Soru, V.Terragni Hybrid Mobile Apps June 1, 2015 8 / 21

Study Design Data Extraction

Objects and VariablesVariables (2)

permissions Identify the permissions requested by the app (RQ4)

rating The average rating provided by the users(RQ5)

reviewsCount Number of reviewsa (RQ5)

aHigh-quality mobile apps tend to get more reviews in its app lifecycle[DK13]

I.Malavolta, S.Ruberto, T.Soru, V.Terragni Hybrid Mobile Apps June 1, 2015 9 / 21

Study Design Data Extraction

Data ExtractionThe extraction process

I.Malavolta, S.Ruberto, T.Soru, V.Terragni Hybrid Mobile Apps June 1, 2015 10 / 21

Study Design Data Extraction

Data ExtractionExtraction - phases 1 and 2

TotalFinance 463Medical 490Transportation 438Travel&Local 484Health&Fitness 352Libraries&Demo 418Business 488Lifestyle 497Social 491Sports 497Shopping 487Education 493Book&References 472

TotalCommunication 487Entertainment 487News&Magazines 491Comics 465Weather 495Media&Video 483Productivity 495Photography 494Music&Audio 477Tools 489Game 492Personalization 493

ALL 11,917

Replication package with complete apps list downloadable from:http://cs.gssi.infn.it/hybrid_googleplay_analysis

I.Malavolta, S.Ruberto, T.Soru, V.Terragni Hybrid Mobile Apps June 1, 2015 11 / 21

Study Design Data Extraction

Data ExtractionExtraction - phase 3

Open source tool to analyse the APK and extract:type, frameworkName, libraries, permissions andsize.Example: identify Apache Cordova

The tool checks if the APK contains:

• an XML file in the path/res/xml/config.xml

• string “org.apache.cordova” in that XML

• a Java class called CordovaActivity.

Tool publicly available athttps://github.com/GabMar/ApkCategoryChecker

I.Malavolta, S.Ruberto, T.Soru, V.Terragni Hybrid Mobile Apps June 1, 2015 12 / 21

Findings

Findings - RQ1

Presence in the Google Play Store

Finding

3.73 % hybrid mobile apps (445 over 11917)

• significantly uncommon among the top-500 apps within 25 GooglePlay categories

• hybrid apps are infrequent in categories requiring closer interactionwith the Android platform and hardware: Photography, Music, Audio,Tools, Game and Personalization

Note

Clear indicator of a future area of work for developers and vendors ofhybrid development frameworks

I.Malavolta, S.Ruberto, T.Soru, V.Terragni Hybrid Mobile Apps June 1, 2015 13 / 21

Findings

Findings - RQ2

Used hybrid development frameworks

Finding

Clear winners: Apache Cordova (258) and Appcelerator Titanium (116)

• other frameworks are less used

• in line with informal claims in other researchpapers [MCM+13, OT12, CSS12]

Note

Appcelerator Titanium has a spike in the Finance category. It has specificsupport for enterprise-level authentication and authorization

I.Malavolta, S.Ruberto, T.Soru, V.Terragni Hybrid Mobile Apps June 1, 2015 14 / 21

Findings

Findings - RQ3

Third-party web libraries for hybrid mobile apps

# Web library #apps # Web library #apps

1 jQuery 267 11 Underscore 29

2 jQuery Mobile 106 12 Backbone 27

3 Json2 99 13 Jasmine 27

4 Ionic 58 14 Lo-Dash 21

5 AngularJS 55 15 RequireJS 21

6 Google Analytics 38 16 Bootstrap 20

7 Fastclick 35 17 Mobiscroll 20

8 jQuery UI 32 18 Crypto-js 16

9 Moment.js 32 19 Datejs 15

10 Facebook SDK 30 20 TweenJS 14

I.Malavolta, S.Ruberto, T.Soru, V.Terragni Hybrid Mobile Apps June 1, 2015 15 / 21

Findings

Findings - RQ3...continue

Note

• 16 out of those 20 top used libraries are not specific to mobilebrowsers. Developers reuse and build on JavaScript frameworksalready existing in the desktop web area

• Among the top-20 libraries there are two MVC JavaScriptframeworks, namely AngularJS and Backbone

I.Malavolta, S.Ruberto, T.Soru, V.Terragni Hybrid Mobile Apps June 1, 2015 16 / 21

Findings

Findings - RQ4

Integration to the Android platform and other apps

Finding

hybrid mobile apps are generally in line with native mobile apps

• 17 out of the top 20 permissions are requested to the Androidplatform, whereas the remaining 3 permissions are requested toGoogle services

I access the Internet and the network connection stateI write to an external storage drive (e.g. an SD card)I geographic position of the user

I.Malavolta, S.Ruberto, T.Soru, V.Terragni Hybrid Mobile Apps June 1, 2015 17 / 21

Findings

Findings - RQ5

Perceived value

Finding

The average of end user ratings is: for hybrid 3.75 and for native 3.35

Finding

Native apps have been reviewed in average 6.5 times more than hybridmobile apps

• Hybrid and native mobile apps are performing equally with respectto end users’ star-rating across all categories.

• We can interpret the review count disparity as an indication of thefact that hybrid mobile apps are neither perceived as too satisfyingnor dissatisfying . (see [HZP09])

I.Malavolta, S.Ruberto, T.Soru, V.Terragni Hybrid Mobile Apps June 1, 2015 18 / 21

ConclusionandFutureworksPerceivedValues

FutureworksPerceivedValues

Empiricalstudyfocussingontheenduserperceptionofhybridvsnativemobileapps

•intheperceivedperformance

•intheperceivedbugginess

•intheinitialdownloadoverhead(appsize)

Empiricalstudyinvestigatinghowhybridmobileappsperformacrossdi↵erentplatforms

Surveytargetinghybridappdevelopers

•focusonpractitioners’perceivedstrengths,limitationsandneeds

I.Malavolta,S.Ruberto,T.Soru,V.TerragniHybridMobileAppsJune1,201519/21

Conclusion and Future works Conclusion

Conclusion

I.Malavolta, S.Ruberto, T.Soru, V.Terragni Hybrid Mobile Apps June 1, 2015 20 / 21

Conclusion and Future works Conclusion

[CSS12] Luis Corral, Alberto Sillitti, and Giancarlo Succi.Mobile multiplatform development: An experiment for performance analysis.Procedia Computer Science, 10:736–743, 2012.

[DK13] Diya Datta and Sangaralingam Kajanan.Do app launch times impact their subsequent commercial success? an analytical approach.In Cloud Computing and Big Data (CloudCom-Asia), 2013 International Conference on, pages 205–210. IEEE,2013.

[HZP09] Nan Hu, Jie Zhang, and Paul A Pavlou.Overcoming the j-shaped distribution of product reviews.Communications of the ACM, 52(10):144–147, 2009.

[MCM+13] Emiliano Masi, Giovanni Cantone, Manuel Mastrofini, Giuseppe Calavaro, and Paolo Subiaco.Mobile apps development: A framework for technology decision making.In Mobile Computing, Applications, and Services, pages 64–79. Springer, 2013.

[OT12] Julian Ohrt and Volker Turau.Cross-platform development tools for smartphone applications.Computer, (9):72–79, 2012.

I.Malavolta, S.Ruberto, T.Soru, V.Terragni Hybrid Mobile Apps June 1, 2015 21 / 21