Hitotsubashi University Repository

TitleEffectiveness of KWIC Index as an Information

Retrieval Technique for Social Sciences

Author(s) Matsuda, Yoshiro; Matsui, Sachiko

Citation Hitotsubashi Journal of Economics, 15(2): 15-40

Issue Date 1975-02

Type Departmental Bulletin Paper

Text Version publisher

URL http://doi.org/10.15057/7996

Right

1

EFFECTIVENESS OF KWIC INDEX AS AN INFORMATION RETRIEVAL TECHNIQUE FOR SOCIAL SCIENCES*

By YOSHIRO MATSUDA**

AND SACHIKO MATsru***

Introduction

The literature search is one of the major aspects of the research activity and deserves

the researchers' serious consideration and yet has attracted least attention except that of

the documentalists and librarians. But recent increase of the publications and growih

of the various means of information have brought up the mechanized information process-ing in literature search first in the field of natural sciences and gradually in the field of social

sciences.1

Recent development of the computer has made a revolutionary change to the process-

ing of the non-numerical information. The powerfulness is overlooked or not fully realized

among the researchers in the field of social sciences ; it is so even among the econometricians

whose daily work heavily depends on computer and even more among the historians of economic and social thoughts whose hostility against computerized information retrieval

is quite strong.

Our survey, though limited in scope, shows that several important trials of com-puterized infonnation retrieval in action in the field of social sciences have been ventured

mostly by the social scientists of particular disciplines without or with least collaboration

of librarians.2 Thus, the mechanized literature search is faced with the indifference both

from users and librarians. This paper aims to fill the credibility gap between the researchers

of documentation and researchers of social sciences through presenting the main findings

~ This work was carried out by the group conducted by Yoshiro Matsuda with the aids of the grants from Isetan Scholarship Foundation for 1970-71 and the Ministry of Education for 1 972-73 (Experiments for comparison of effectiveness of retrieval by KWIC index and facet classification scheme using the titles and abstracts.) The authors wish to thank to all the participants to the project for their stimulating dis-cussions and co-operation and for their generosity to use whole results and findings freely in this paper ;

Professor Jun'ichi Akiba of Hitotsubashi University, Assistant Professors Shigeyuki Sato of Hokkaido Uni-versity and Ryo Suzuki of Saga University and Lecturers Akira Hirota of Hosei University and Kiyonari Kidahashi of Hokkai Gakuen University. Computer processing is executed at Computing Centre of Hok-kaido University. ** Assistant Professor (Jokyo~,ju). Documentation Centre for Japanese Economic Statistics, Institute of

Economic Research, Hitotsubashi University. #* Assistant. Department of Management Sciences, Faculty of Commerce, Otaru University of Commerce.

1 Janda [10] especially Part 11 ; Research applications, Matsuda [14]. s More exactly without the initiative of traditional librarians or bibliographers. See the Table in Ap-

pendix I .

,

16 HITOTSUBASHI JOURNAL OF ECONOMICS [February

of our trials published in the KWIC index series for social sciences,3 comparing them with

other similar trials in the field of social sciences.

Such advanced documentation activities undertaken in a specified discipline as ab-

stract and / or index making have often encountered with suspicion against their usefulness.

Such suspicion shares the following two suppositions.

i) In the case of social sciences, Iiterature search is so unique and peculiar to the

researcher's personal frame of reference or value premises that any information pro-

cessing planned for common users is quite unpractical for the researcher. And librarians are expected to remain as a neutral interpreter to writers and users. This

"librarians' neutrality" is guaranteed so far as he depends on usual devices of infor-

mation control such as classified catalog or subject catalogs in a traditional form.

ii) It is for the current awareness services and not for the retrospective search which

is indispensable for social sciences depending on the search of so huge pile of volumes

stored from the past that the recently developed information retrieval techniques are

most suitable.

The first proposition is less supportable judging from the various surveys which show

that the librarians' classified and subject catalog themselves have come to be a jargon not

only for the outsider researchers in other desciplines but also for the insiders like library

science specialists or information officers.4

Severe attack comes from the researchers in the interdisciplinary sciences. For "the

librarians' neutrality" is supported by the bias that the librarian's or indexer's judgement

is itself neutral. But such neutrality can scarcely be supported; because the descriptors

used by librarians, which are so to speak subject headings, require the writer to transform

his intention into librarians' vocabulary and the user of these descriptors is in turn requested

to transform his retrieving vocabulary into librarians' vocabulary. These double trans-

formations will not guarantee the equivalence of the semantics of the writer's descriptors

and user's descriptors and make it difficult for librarians to remain as a neutral interpreter

or translater.

It is clear that in the case of classified catalog the same problem occurs more severely.

It should be well remembered that Bliss classification group and CRG inspired by Ran-ganathan's theory of faceted classification have gained practical success in particular subject

area.5 This suggests that universal classification scheme is difficult to be designed and

single purpose classification scheme is much more powerful and, thus it may be implicitly

supposed that librarians' non-neutrality or the homogeneity among the writers, Iibrarians

or idexers and users. From this it will be well supposed that librarians' efforts could be

flourished when they commit in a particular discipline or a frame of reference. It is this

kind of reasoning that stimulates the direct descriptor making tendencies represented by

the KwrC or KWOC indexes now in fashion.6

3 See [B-6]. 4 Hans (Hanan) Wellisch ; "Subject retrieval in the seventies-methods problems, prospects," in Well-

isch [24] p. 16. 5 T.D. Wilson; "The work of the British Classification Research Group," in Wellisch [24]. They still

hold the possibility of compiling the effective universal classification scheme. 6 As to the development of these techniques, see Fischer [7]. Another approach of compiling descriptors

･eliminating the indexer's bias, see Jean Aitchison ; Thesaurofacet ; a new concept in subject retrieval schemes," in Wellisch [24].

1975] EFFECTIVE)~TESS OF Kwlc INDEX As AN INroRMATION RETRIEVAL TECHNIQUE FOR SOCIAL SCIENcEs 17

However the mechanical information processing is undertaken in most cases through

the collaboration of the researchers of a particular discipline and documentalists or com-

puter specialists, excluding the librarian's participation. These facts show that a philoso-

phical change within library science is required even when it is applied to the field of social

sciences. Thus we should examine the plausibility of the second proposition.

Table I shows the chronological development of KWIC index applied to information retrieval in social sciences and natural sciences. The time lag of the impact of computeri-

zation is not large between them but the stage of practical application comes so soon in

the field of natural sciences while that in social sciences is staggering in the experimental

stage.7

But, even from this short table, it is clear that the KWIC index compilation is more

favoured in the interdisciplinary sciences like sociology, cultural anthropology, political

science, econometrics and computer sciences or in the field where like legal processes

immediate informations are required. Thus this kind of information retrieval techniques

has a bright future in the interdisciplinary sciences. The most important point is that

trials in the social sciences are mostly oriented to the retrospective research and not for

current awareness service. The prosperity of the cumulative indexes may be its evidence

and acts also as a counter example for the first proposition.

Needless to say there remains a supporting evidence for the second proposition. The

past publication of mankind is so huge that even wealthy United States cannot afford enough

financial support for her RECON project ; to convert her national union catalogue into

machine readable form. But there are two examples qualified enough to support the retrospective search in social sciences. One is Harvard University's Widener Library Shelf-

list Conversion and Publication Program resulted in the publication of the Harvard Widener

Library Shelflists which are facsimile of computer outputs having author, title and chrono-

logical lists in addition to the main list (classified shelflist)8. This series promises us the

future fertility of the retrospective research brought out through the further corporation

of the important libraries sharing the MARC (machine readable cataloging) program.

The second example is Pollin's Godwin Criticism which covers about 3,379 items published in 1783-1966.9 Although it has some defects which we will discuss later, it has

broken a new path for the retrospective search by machine. And we might be permitted

to register our work as the third example in this field. It deals with about 686 works of

Claude Henri de Saint-Simon and Saint-Simonians published in 1817-1971.

I. Natural Language or Descriptor

Here we will analyze the possibility of neutral transformation of the contents by

virtue of the librarians' descriptors called Subject Headings (hereafter often abbreviated

' The cease of publication of Jurindex occurred in the second year of its short life is quite symbolic and it symbolizes the present situation of the mechanical processing in the field of social sciences. A few peri-

odical publication of the mechanized indexes are established like the Annual supplements to the Universal Reference System; Po!itical Science, Government & Public Po[icy: an' annotated and intensively indexed com-pi[ation of signlficant books, pamphlets, and selected and processed by the Universal Reference System-a com-

puterized information retrieval service in the social and behavioral science. 1 965+

' De Gannaro [51. g pollin [B-51-

18

TABLE 1.

HITOISUBAsru

CHRONOLOGY OF THE

JOURNAL OF EcoNoMlcs

DEVELOPMl3NT OF KWIC INDEx

[February

(ABRIDGED)

1975]

1974

EFFECTIVENESS OF KWIC INDEX AS AN INFORMATION RETRIEVAL TECHNIQUE FOR SOCIAL sclENCEs 19

Matsuda, Yoshiro, Sato, Shigeyuki Hirota. Akira & Matsui, Sachiko ; Bibliography ; Works on Saint-Simon and Saint-Simonians. 1831-1970. Vol. I.

Matsuda, Yoshiro et al.

Bibliography ; Works on Library and In-formation Science. 2 vols.

as SH). They are expected to be called neutral to both writers and users but in fact, should

be regarded as to be subject to the librarians' frame of reference or paradigm. The most

important paradigm of librarian to compile SH is "public use first, specialist use second".

Thus users sometimes encounter with a serious gap between technical terms which now become of their common usage and SH terms using colloquial and sometimes unsophisti-

cated words which are seldom used in researchers' discussion. This fact must have been

stemmed from the belief that the paraphrasing by daily language will help the recognition

of the public easier. This belief is severely criticized by John M. Christ who takes the SH

of the Library of Congress of the United States as an example of the most popular SH.ro

He compared the subject headings of social sciences in the 7th edition and the Sup-

plements of the Library of Congress Subject Headings (hereafter abbreviated as LCSH) with

TABLE 2. THE DESCREPANCY OF TH1~ UsAGES OF TECHNICAL TERMS AMoNG LIBRARY SCIE~ISTS AND SOCIAL SCIENTISTS*

* Figures are based on J.M. Christ's survey. See Christ [3], pp. 68, 83 & 123. A miscalculation of percentage is corrected. Comparison is based on the Subject Headings of Library of Congress (LCHS). 7th edition. [22].

** IESS is International Encyc!opedia of the Social Sciences, ed. by D. L. Sills, New York, Macmillan and Free Press, 1968. *** Interdisciplinary terms are based on the terms used in the 13 basic interdisciplinary social science

texts. See Christ [3], pp. 7,~5. Whole list of these 128 terms are listed in table 5 of Christ [3], pp. 76-81.

**** Sociological terms here defined are extracted from the indexes of Broom Leonard & Philip Selznick's Sociology ; a test with adapted readin~s, Evanston, I11., 1955 (287 terms) and Robert Faris (ed.)'s Handbook of Modern Sociology, Chicago, 1964 (235 terms), sharing only 41 terms in common.

lo Christ [3]. As to the criticism and defence to the LCSH, see Richard S. Angell; "Library of Congress Subject Headings-Review and Forecast," in Wellisch [24].

20 HITOTSUBAsm JOURNAL OF ECONOMlcs [February

the index headings from the International Encyclopedia of the Social Sciences. His com-

parison was carried out as to i) general social science terms, il) interdisciplinary social

science terms and ili) sociology terms and other detailed comparison of particular concept

like "value". His conclusion summarized in Table 2 was that LCSH has governed by different frame of reference from that of the social scientists. A quarter of general tenns

cannot be traced and in the case of sub-terms having the form of compound terms un-

traceable ones amount to 93.3~･ The LCSH prefers single word heading contrary to the tendency among the scientists to prefer corporate words or collective terms,u This is

shown in Table 3,

TABLE 3. CHARACTERISTICS OF THE SUBJECT HEADINGS SHOWN IN THE NuMl3ER OF WORDS IN THE SUBJECT HEAI)INGS

Terms without subject heading referents

Terms with subject heading referents

Sources: From 1) to 3) figures are obtained from Christ [31 pp. 68, 69-70, 83-85, 121-3. But some counting errors in the original tables are corrected.

From 4) to 5) figures are obtained from Matsui [15].

To interpret his finding we must take into consideration the following : that articles or

index terms in Encyclopedia treat narrow concepts like articles in academic journals or

in one chapter of a book but that books on which LCSH is assigned are apt to treat much

broader concepts. But his findings are still striking enough. And this must be the back-

ground of many trials of thesaurus construction.

** LCSH is also slowly catching up this process. ~ In the 5th edition of LCSH one thirds of the LCSH are single word and the 7th edition shows increase of corporate or multi words. See Daily [4] p. 3962. And the Table 3 shows that this tendency is still going on after the 7th edition.

21EFFEC皿vENESS　OF　KwIC　INDEx　AS　AN　INFORMAT【ON　RETRIEVAL　TECHNIQUE　FOR　SOCLへL　SCIENCES1975

O，

頃内，

0つ一

～σ

（つ

一一〇

ひ晒貯

一一1，騨・●A－〇一1∩‘9一

ひO∩一9一

円－こ　1凶

7¢一　1【

Oo∩ー（

OO一8一

40（　o～

～（ハー（

内ヨコ嗣一

ρぴρ9ξ

へ6一　畢↓

0トハ　陰～

O～～一－一

〇ひ　－一

、圃o　I（

　タ噴P

　nOOP

　　　嵩の

　　　偲窃

∩，Oひト2

円10ひト再卜

一IO伊魅ト

ぐ〇一ひト｛一

n8一，卜雨卜

　　卜㈹Onト

　　OOトのト

　トOOn一

　　　　　〇

1ぐ～一

lO四一

1∩一」

oξ～一

90　一

〇一n』

ー‘　」

9ぜ　一

〇〇～U

oO　J

量ひNU

よ（ロ

‘云一

－可　一

〇〇～山

，卜∩一

60（吋

14　」

－尋～四

9つ6］

胸訳㊦

2名

卜心いn

“（（n

Ogn

ひ一ひn

。o

n

O一ひn

ひOひn

貯ぐ⑩n

①一トの

四両ひn

い一訳

之一～

～一マ内

笛ρ“へ

‘卜罠

∩一ξ～

Gスヘ

いひぞN

馬（」“

へトnへ

OOρ内

の～4内

一芝，～

象卜～

卜O（梶

4『O“

00卜～

OOO～

0へO～

寸h一内

の晒トヘ

』Oρεコ属⊆』Oり一

匡αφ匡PO）20⊃O鷺．（N一ΣO一9一一年の⊃OZ一コ）．卜ε一肛W荏）〕」〇一コO山UZ（国砿

二岩卜OZ一〇の↑Z］Σ一Z2一のZUOZじ．ヌ。一崔望一〉OZ嵩旨ひ一の岩田）Z（拐

OU¢く」匡け一一記）（L四〇一一、OOα⊃O匹一Z（＞（の］⊃里一一〇低一囚の国⊃∂一」ζZ一一∪の

　　　　　　　　　　　蘭．（OO⑩一－卜Oqo一V．田」）一一の　］［］一＞コ一Z畳×一〇　⊃O　の田⊃O一匹一トZ一一∪の

四工O卜，餌コ〕h）POO匡L］ご．（匡巴工（）」⊃〕のZ⊃》済卜匡《5y⊃（］口｝の一〇一］㊨田）Z巴）の

］⊃3一〇〇一工Uα》．（ぐ陰∩　一Σ0一．匡⊃U一）⊃OO肛“　一一）．g㌧戸¢（IX⊃《一〇　のUO　一U　の吋）Z一－∪の

Uつ貿OO左国3．aー一］70㌘雪冒）⊃OO窟低U一）あ↑宴妄コ（］㎝の岩旨器uZヒ∪の

〕⊃¢一〇〇をぎ》尋需OZ，］一ZZ（〕卜：NaOZ，認ZZ（澤．］≡窃⊃〇三1一）Z巴）釦q

U⊃3一〇〇田一5．一晒一〇Z．］一ZZ（U卜OIO｛OZ．国UZZ（ω卜。］一¢一の⊃〇三1．国）Z］一∪の

く・一UOOU江一《卜Zり［（一の四一　のに⊃U一Pu田κUの一一匡（“U［」O⊃α．］一（一）Ou㌦　一）Z］一）の

　　　　　　　麹・匡国一αPOL　のU一に（工）　一〇　田一にO田工一”NnO一　一］ZZ（・国」（閏）Oの　］）Z］一）の

OあZ巴ZO三の8一ZぞののコOZ一益にユの寝ぞZZOεコ寝X⊃（］ゴ国＞⊃OZ一uZ巴）切

ト一　一一）　働．（049の一工〉⊃一〇》．一のコω一〇ロ一b匡　国一の〇一〇一の＞エユ．UΣ＝Oエ・」　国O　一）Z一『）頃

」

　（一　（，田一一一』（一．7　匡（ユ．の］↑⊃匹］O　の］O　国『のX（工）　（一　（　田）ZO7鴫0

9国」」一Σ（」（一〕Oロ一（配UZ一〇ZO≡⊃Uに．国一《鑑O翼、噸巴Oエ一〇．一匡田

。のU⊃e一工色O匂りOJ円エ巳×⊃）（匡ト．（］一hエ（“　回Σロn）JU｝匡一の⊃OZ一　］ΣU一

　×つ（）（匡卜　X⊃《　ZO一ト∪コOO広↑Z一、

　のUO9一一醒一の⊃OZ【響」　］O　U⊃O｝エユOのOJ「＝色　」（Z配⊃Oう．匡⊃］一）⊃OO配

の田O．田量一り弓コOZ＝」切O”⊃e≡ユOのO」剛工α」（Z匡⊃Oう匠⊃］一）⊃OO匡

切口O．巴傑トの⊃OZ一二〕〇一（Zエ⊃O？匡⊃り一）⊃OOα且一一、あ．一三（の

巳70一〇一，一一鷹．］ζ臼…OΣ房ー一三（のZO6一」Uα（」ロO」（Z肛⊃⊃「畠

19ZO閂O一一国匡，ZOエあ1卜Zマの〕O］Zを一）OO（」田〇一（Z匡⊃OつU〉

　国O　一一　叫⊃〇一一國一〇店］Oロ一「く匡卜、，ト一〇．く一一　Uり弓『匡一）．」。彰⊃田』⊃

　（一　匡⊃の　田UO⊃一］、

　（一　UO　一一α幽）の一〇　Z⊃甲O　国工一一一」、O　ト”・監匡コ］一《の榊Z（O工O，」8　国

．Z＝三」Z］匡］窃Oまに（α・密⊃P⊃な］這］の畏吐垂嵩（氏－当一山．謹一

の（虚つに亀＞エO』ニ↑」OZC＝）Z《」］エ一」O》O⊃トの（”ZO＝《Z《」OX一）ご＝Z巴）の

　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　■．＞エ］）OUの一〇　）一ら周一Z㌧一）の

一⊃ユ30ート⊇塗く　　o●！エ］J¢Oエ」」《）＝の＝《旨ZC」）Zぎ一」ZOり巳」＝Z巴しの

　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　”，の）一1OZO）U一のU〕Z］一）の

XOO工匹匡《山三J8ZOZOZ《ユ《一Zコ三　の］一〇⊃トの、　　　　　　　巳．～。の］）Z一『）の

エOO工亀庄（」ZコZOZOZ（匡（しZコ三　の里O⊃一の、　　　　　　　　！N∂の〕）Z土）脇

　　　　略・Σ⊃「のO江Σ》の　O配O」Z《hの　トの『口」　一工ト　」O　のOZ陶〇一〕）O匡し9ひ“ひ一9みU）Z一一∪の

　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　8．φ］Σ⊃一〇）　ト一。の〕）Z〕一）の

一帥三〇Z（翠］ヌ”¢（ρ（）E］エ（」OのU↑《ドのO」＝Zコ国エド三の）㌧の吋）Z国一）の

O一の　」0　の一Z一＝一一］　吋エ一、　　　　　　　　　　　　　　　　　　　　　　　　　　　　巳・の」）7』〕噌）の

　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　ー。の囚）ZW一）『

　」（∠O＝《＝缶一三田乞一Z］）一工↑トのOω≡田のく三里⊃一〇＞〇三工じの〕）Z忠）の

』（工四＝卜（Σ　巳ト一二一し、　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　o

マ6Zつ）一Z一の＞《のの国に］＝卜OOZ（・＞OO」O（一〇エ一宙工OZ（ー．3巴O⊃窃

）⊃O（をト三。一・一〇〉　●裂6…OZO）］』の宙のOα×］、　　㌧の一⊃e≡CZO）］

　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　8．切＞（切のU　α切工一〇　〇Z（

（Σ一＝一（芝　ZO　Σ⊃一のOαΣ＞の　＞一」〕どに一』　工一し噂」　一量．¢ヤニα］〕Z一〇Z四　〇Z（

ビユOエのZ【O「　］エじ　ZO＝《XαOし一工O）Z一〇Z《の巳。a。OZ．∩O

O⊃一の《”どO＝《Z《」江ズ］）こ二ZU一）の、

の巴工国69 。

の］）ZU「）の

　の一）Z］『）の

　の］）Z駕「）の

　の」）Z4一しの

　力由UZ一【）の

　　轡）ヱ占閂）φ

一．一）Z一『）の

5。］）こ］一）の

口E5国OQハ》一≧〉【

、しO

（　」〇

一（一）Oの

」≦UOの

一《一）O佑

」（一uOの

」≦）Oの

」（一）Oψ

冒．o唱．導的］葺o旨』

）一GC一　Uエト、

　　のOZ【O醤］）C＝こ．」ハ⊃

臨O

］エ一

」■ト

㌢芦

山＝ト

」工戸

OZ「工∪（」卜

ZLZZZ－O一一一

N

■．一）Z」帽）の　，二　　；（一　〇Z（　＞一輯」一

一（こ⊃一（Z　一■一　〇一

一（『）Oの　」1一　」O

Z（Σ⊃工　一＝ト　Z一

トZ一Σ］¢（Z（Σ　OZ《

一く一）Oの　OZ⊃’　Z［

N⊃く　φ一（）Z（にし

J（一）Oの　」■一　Z一

」（）一“＞1ユ”（　四Σ二JO〉・》一一

一（）闇一一一〇色　OZ（

2「　奮《一　〇Z（　＞卜口」「こくPO窪亀，＞匡〇一工一

）一工OZO）」　凶O　」ユ⊃一）⊃エ↑の

　　　　　　　OZ（　Zつ瞬ト（」幅ユΣO）　窪「占工卜”の叫」カ《一

　　　　　》ト一ハ竃占）輔Z⊃　一工一、

　の］口O⊃一の　」く∪「一（Σ」＝一（Σ　OユOもZ（トの．OZ『＝Σ《

の巴O⊃窃’《）一建＝」」《エO匡O巴Z（トつ〇三＝Σ（

　のOO工一一Σ　」（）噌一（工一工一くΣ、

≦O」ユ〇一）》」Z一だZO二三を一㌧二、

　ZO「hく一7しΣ⊃）OO・工）エ（〕の一＝　しO　ZO一ト（一U記こー

　　のZO一一（）『」ユこ（　工一一㌧　の一のの」）つエユ　）一一の《工）

　＞OO一〇〇〇■』一Σ、

OZ一〇一一⊃①　」一〇〇Σ、　の田Σ］」沿一）エユ　の」O　駒O⊃一湖oO

　　一（一）Oの　一エド　Z一　のZO一一（）一」ユユ（　＝一一雪　の）噂

　　ひN5こ。Z（エ一】ど《7そエつ、」OにOZつエZTの）

のエ⊃」一Z一〇Z『の一3一配01巳く．’乙一《¢由Z四〇ZO二

　の）［ト（Σ栂工一《Σ、

　　　　　　　　ごの（エつ工低　OZ（　の）一一の一一く一ゐ　」（∪一一

　一（）一工O一ψ「工　～鱒　の一『O⊃戸一　＞一噌のエ一）「Z⊃　のッニ

　　　　　　　　　　　　　　　　占O　Zつ『一）Z（」　一＝ト　」O　＞

　　　　　　　　一エP、

（書目層εβ＆ぢo昏）［【

×国OZ一Q一3呂山Ooり…【日山Σ＜×国

．一．O起

．o目．導的］目o』匹

一

lFebruauH『rOTSUBASHI　JOURNAL　OF　ECONOMICS

　（O口∫

OOひ＝

OσO“ト

O（一ぐ一

　ひO卜露

　卜Oトタ

　ひO内3

　の“ひB

　Oξ｛3

内内Oハト

NqO（一

　ぴ3のタ

　一（コ3

　へ内Oヌ

　　O（O

　内ひひヌ

　ト一N3

　一つ【3

　ハO。，3

　ぞト一ヌ

　ひξむヌ

　ひO内ヌ

　σり一ヌ

O内〇一3

　　一内O

ひ¢Oξ

　一ひ03

卜｛へ郵

一【〇一葺

oひ一トo苛網oo“～o卜⑩ひずトoぐξo伽ド、⑳oひぐ四〇ひOOε～ひrOひξ・0・0りトGP内nA‘3ハ∩『寸的ひ個∩曹内Nぐ一〇一〇向一〇〇一〇一〇〇一c噂一一一〇〇一一〇一一〇〇〇一n∩nnnnnnnnnnmnn〔n∩nA∩nnnハnnn∩

3訳O内㈱＆糊

一嘘QO

O四卜一

〇300

0300

～叉》O

Oめ酎一

〇h～網

い門四一

～3N一

絢茜一

〇3QO

Q5り〇一

〇300

08Q

O3QO

O700

0300

ト一〇一

一二〇

冒コ

卜〕

］毘嚢J望

（恥脚い一，］9閣α）

（．N　菱旧一4一3

（切　国　一一く『ワ…　．呈一鑑》h

同　］喜一の、　　●の　髪

トト⊃りV　一〇しり喜隅』陶

OU　．3　　・の喜一」

（雲覆．三量崔）

垂卜塁）　1（N　．2

工Q、　，＞U翼」）　1（Oξ一8卜3一

∪の　》塁ロ一一　」Q　望冨

　　　（．N　【ζ一JJ一3

浸⊆工　，の同箋）　一（03一IU3一

峯弓ノσα峯U）

　のく0　喫コ」　ト」《ト窪

8と言δ錺一舞）h8） 　

●］ぎ《島U　●窒］hゴ一一　．」跡一⊆Zqこ　・し駒弓噸CG」コ賦　．妻UJJく

　露。毘国ト『堅QU駐く　5同一琶を一　，ω握《婁（一　〇男ム一一§J《

（．］Z一髪　．］ロのJ⊃くト）　一（“旧一ーりり一．α一．）切一一］）窒ト陶α8一《

　．一¢塗）一（3U一13り一．低一。〉o堕一一〕）＝一　切］一髪配m一」　，（幅愚一ζ

　　　　（．促　．く　9》一臣簑Qロ⑳）　o（口ひ10，①．底n．》の同一］V　セ》《配q5冒」（

　　　　　　　　　　（ひUひ輯・（〉Oと一ンQと）　魯．（¢QJζトくと婁ト一》（」J（

　　　　一〇　くU一）胃り　（∪］一〇一Jの一口）　　．国トト誉の　¢巴　Oり陶トロロく」」《

　頃」ぎ》国工ト5」匂り］〉一↑5覆U卜Jく　一U一q5くトq］8QIU一トUロく」一く

U壁トエ一一38h一【」QくくO歴〕一」く　一］」の⊂卜配臣Q塗U一一Uの《」」《

］翼一ヱト一30］ト』鱒」Q遵O距U一」ζ　悼U一ロく一肛UQ51∪一ト一〇く」’（

　　　　　　o（nり一⊥四マ一。巳　“．）脇h一］）　UU一～U〉　，ω⊃一一望（＝もら⊃〇一《

　．

］9QのO］卜臣一りO⊇一」幅　、ζQトのQ＝【と＝↑⊇Qの、＝コUの⊇義h配切の一（

　　　　．氏　一．）切一」〕）　U裟U因りのト嚢⑳隔」」9茎∪しら　．》萎00J（

　　　　o窪Oゴトく差Qト3思」Qトト肖塁U》肉～コ］ト（h切　．》奮一《

　．一¢窪）　■（ゆ3一〇喝3一，し甲．〉oら一」〕）Z一〇Uh髪O一一　pq一菱①」《

　　　　　。こ一．）の一」］V；Q騎一く【UO切撃＞α奮一一一一〔一の蔓匂り（」く

　　，（U7一Ih3一．巳　一．）の同JUV乙【トく国》QUりのく＞髪①【」隻く①q一（

＝四工Uの一工塵トビeコ～里ロ．卜」く工∪切一一U駒0冒ZU工∪の一工§と曙

　　薯ロロ．りり恥一　く一匡①（H窪の佃13四　・¢のののと⊇薯h幽＝崔とく

U　Zりヱ一の0卜Z一《uり、

《の、

J《鴨）Oの　ZO一ト）つ署トのZQ）副広

」．ユ　］O　X⊃《》くユh　の］一　ト］

U）一トOZ、

UOの一瞳

］一＞　《」

X一〇Z

u

、、、、、、

．国望（薯

p望賢塁

9Uoら口くぎ」の］コ窪」の］」髪　の翼〕一雲《

”U呈］圏りO　ユU卜婁》りOト　5【ト登』』乙一、

U一α一くに

と《U－600、

（一一の』話一Z《馨　一　《肖一誤もり一∩δ㊤円～と　》〉望8、

QψQ一⊂ト（Uぎ　⊂O一コO、　　　　　　　　　一（一　9《一」琶Q凶一①嗣ロ　一口　【QO

駒op碇U↑トコリ　。（　．∪、ト臣一」ωエト」02ロト望円　OU雪ωooOト◎りど一一一Z

脇，崖一ト一コ∪　．ζ・．U、　　　　　　　　　　　　　　　　　　　　　　（Ψひ〇一．5

の，“］hトコ∪　．く　．∪、　　　　　　　　　　　　　　　　（．O。Z・匡U一トコり〉　5。

璽《　《叫さ一U一＞）　　臼ン埋　曾O〕　．四　　．窪一2一の善①￥　　　　　（一噂〇一．

ト《　￥S㌔　，里　旨　》卜【臣］）一Zコ　］卜（一切　．墓」司、　　　　　　　　（贋「　Z

　　　　　　　（。うぎ「　．＞回崔」）　8（①3一，卜7一・ユ　一・ンOhJ］）　］暑U一

、Zぎ一一」隅；一埋」¢岩　駐U瞬α〕のoり壁《、

一とU一〇幽Joo一①　N困切トZ型Uきと　．工り⊃《Z

3とoII

臣壽

艶嚢…

…園

り

穂∫象

1輔

（o

ε甥Q雨唱り5山一500守暑O】、図＜∩O出）［寸

　　　　　　　　　　　　　　　　　　圓．山⊃9一

　　　　　　　　　　　　ロ．の切⊃9『一一一〇ユ

（切］⊃O「■OZO）』　ト一　のし⊃e一ト一JO巳

．（W⊃3一工OどO）］　トU　の国⊃G【一『JQユ

電嗣。（u弓〕⊃O［ト一JCユ　一国　の〕」く竃Q工

　置．（の］，〇一h剛JO巳　ト国　の一一（¢O工

　　騒．（一⊃9｝卜『一〇こ　トし　uり一」（匡O工

5コ注さ冨雪冨召≦器8ら竃ユ≦七8缶3・（ま（］隻缶一

一くの］O一］ZU肇O・O．匡〕一工⊃O」国OハU＝〕帥内》のの〕」蟹⊃の一］ρ一」く一）Oの

に低．Uα隅《卜土）りφ］刷¢9Uエ一《」田OUロ国¢O（zコトz《zU一zOu・］」（一）oの

一ト《エし一切＞“　一Z一］　．エ¢O」］工一《脚NOの配」C切O《一〇Z⊃記OOZ⊃　」一『N

　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　巳．n．一〇）　　曹

u一）一エα08’こ…累』φ一（切zoこ《ごζこ『2（52雲。）

　　　　　　　　0．Q卜Z］工一〇肛Oの『¢　」一〇〇Zく一Jくト一　〇¢〕一切‘一ユ　一一Z

　　　　　　　　　　　　　　　　　　　　　　　　　o．の一ヱ一一〕J　Jの∪工卜1．〕コe一のつΣ　（一

　　　　　　　　　　　　　　　　　　　　　　　　　　　　　9。国コO「工氏Ogり》J凶τユ　」王UQ氏

　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　一可OO一1）N⑩一　　。ω一¢畷ユ

　　　　　　φ⊃工の一」《一NO㎝　㎝吋O　］トニ）一二∪の田O　ユ⊃～　O（肛b『uの　，二一

　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　■．ZO「陣く一ζ国のの一〇

一N8缶。岩59⊃舌。Z⊃一〕一Nの一吃塁」⊃ニ）の島Z苗の。Z⊃

　　　　　　　　　　　　　　　　　　　　　　　　　　■．X一〇，鞠《工Z《　α国≧一〕の　Oこ⊃

　　　　　　　　　　聯．ZO二（一匡Uのの乙．」こ■くど～国の切《一どZOン』配工一一

りhuOα〕　α叫O匡つト（ζ一P一JZ一工）φ一⊃一〇¢」OZ一Σ四トの＞の

　　　．．σZコZ（」ユのト」《工）のト¢「3　ユN　］肛一（しlN］のの一《一　工ハ）〉

　　　　ロ．O」《NくOOZ（ユ《o一三《の＝O）Z一の国一〇Z⊃Uエエ．コ

⊃　切⊃エの一」《輯NQの　の一〇　］トェ）「エ）U叫O、

　　　　　　　　　　　　　　　　　　　　　　　　　　　2．ZO一一（戸配UOの『O，．

」閣トZ叫町）の

の四）Z一『）の

の」）Z噛一）の

の一）ど」『）の

の〕）Z山一）轟

のω）Z」「）uり

のU）Z泌『）ω

．田）Z］一）の

●U）Z一一）の

〕）Z］一）の

　一）＝一一）い

　　の一喀●＝田」⊃二）㏄

　　　Z］一毬「乞工）♂

　O工の「ZO≧一のZ（の

くZ≦ZO三の4誘

　　　　　一〕　O二4の

　　　　験吋国匡（Σくの

　　　　UOZOごの

のZOΣ一カー一Z「（の

●のZO＝一のー一Z『（の

切ZQ≡のードZ＝の

のZO＝一のー一こ一（の

⑳ZOΣ一の8卜～「（の

U工）の『トの輯ZO工一の1卜Z一《の

　　．ZU一の一ZO工『切曜トZ一（膚

　　”φ⊃工匂り一ZO工噌の1卜Z一峨の

ー，の⊃X2ZO三の5一≡（の

　　勉億⊃ユun【ZO三の置一三“の

　　の⊃ぎヨZO三の1卜≡くの

謬

．O信．℃－螢］∈O』匹

　国Σの一」く【）0の　⊃O　の』の一《）Zくτ」

　の」0　】一〇）一。一　く　叫のの宙匹Oこユ

）一（よQ一）OO　切一　肛⊃Qこ　山の〕▼一ト9・

》一く配O一）CC　」一　1⊃」よ　一の一工戸1・

　一口つ　〕『Σ］O（）く9」　一（一の一v二〇z」＝》

　の』O　］姻Σ山O（）く。一　UO　の一窟「OΣU≧》

　の」O　劇一工一〇《、）く●一　一〇　Xつ（》《γ』

　しC　ZO【ト⊃JO）国之ーα山卜‘⊃」）　却＝ト、

　（」　一』　の」一）く窃一Σ　の］」9⊃し一⊃、　一四

　（」　〕二　一4一U・一　エコ”　」く＝』7一〇　」網｝

　＜一　国O　ωO，ト田oJ　（

　ヱ恥㌣よ国の　OZP　の‘し｝噌の匿トど「くψ

　一一J］乏｝（の一）、

　」一〇、一〇〇）一α　」《　OP∠一ZJJ↓ユつ切

　くZ一ぼ一一QO　く一、

　O配O一ρ￥

　UO　一）巴こO　（一、

　ZD、

　ZW蟹っ一

　Σ」Pの＞の　」＝）の【乞］≧乞O桃四工」く［NGの

Z出9＝］COと⊃」畠跳旨〔】

　］客工』一　一「0　5＝⊃

、　の（O　い　OZく一工）のト⊃一〇　n唱りZ⊃う　OZ⊃

　りおeうZ♀二のo卜三（の、

－UO、

匡国OOZ⊃㍗δ三のー一三くの．四二り》

　乞ψO　（一Z⊃　ZOX一匂n巳一Z胴（U、

酢⊃）Z　工一〇、

寸

　　の］）亡⊃つの　のU一、

のエ⊃9）　．一三の｝ZOて目の6卜記門《の　一一　ト

　㎏Σの「」《「）りい　歯一　ト］　ZりΣ噛のートこ一

　のZ一『70と一ψ『一Z圏くい　の国J　N国＝）　留

どQ一ト）⊃00ぽトZ『、噌一〇，O

　　　　　　　イ一一C一

．一じz㌶《）　】一＞一〇ユ亀一＝　ロよ（りく

．一り～X畷）　国ト》一〇巳氏｝工．ユ　エ⊃の

］とハ一29ζ『切ーヒ・二くの　ω一　よ⊃の、

　目

」　］こ一ど田　一］9く…）Z田　〕の⊇ロ

ハンO　tコO）　こ⊃。O　包一一）〕Xユ，

　　　　　　　氏⊃り）　～⊃－O　UO一）叫

　　窪切（｝　ψイPJJ冨トの亀く（一　］二）の

く）一一つ一トく）　（Z『1ト一UO　（一」く

Z〕＝）一一h」くエ）のZ〕のの『3J《『『つの　ロ一自、

　　　　　　　　　　　　　　　　　　の兵O、

　　　　　　　　　　　　　　一工）の一一くΣψトの距の

　　　　　　　の⊃Σの『ZC王一の8一イ『《ψ　と」O、

［。・

一Z『一　．£工り」，ズ一一《

匂弓

Σの一ZG≧一の1卜Z一くの、く工

Z

．の⊃『の『イ⊃工rりど　Z一ユ」⊃一～　02

．O口．℃－国］口』O』』

6り

22

1975] EFFECTIVENESS OF KWlc INDEx As AN INFORMATION RETRJEVAL TECHNIQUE FOR SOCIAL saENcEs 23

His findings suggest us to use another kind of descriptors for subject retrieval. One

extreme way is to use natural language used in the text, to be retrieved. "Natural language"

used here means vocabulary uncontrolled by indexers or librarians.

The most practical system of information retrieval using natural language may be the

KWIC index system formulated by H.P. Luhn in 1959.12 As is well known KWIC means key-word-in-context index which uses title words as keywords instead of the thesaurus

or SH except some stopwords preassigned and they are listed in alphabetical order in the

middle of the page, accompanying the rest of the title after and before the keyword. See

illustration in Fig. 1.

To compare the KWIC index with LCSH we should use the data base consisting of the bibliographic description of the documents in book form. According to our KWIC index to management sciences summarized in Table 4, the subject headings (LCSH) whose

TABLE 4. DEGREE OF CORRESPONDENCE OF TITLE WORDS To THE SUBJECT

HEADlNGS OF LIBRARY OF CONGRESS

Total different LCSH examined are 558 in 7th ed. and 77 after 7th ed. Some adjustments are carried over in matching the LCSH and title words and they amounts to three percentage of the total subject headings examined. Source: Matsui [15] pp. 295-298.

words and / or phrases share the title words in common are only 68.4 ~･ And D.H. Kraft's

similar survey in the field of legal periodicals shows it 64.4%･13 Although V. Mostecky pointed out that the subject headings should not share the words in common with the title

words in order to avoid the vagueness found in the titles.14 But Christ's figures shown in

Table 2 indicate that through the subject headings almost 54 ~ of the technical terms get

no direct access and they must have escaped from Mostecky's blame because they are more

exact than the title words and belong to the researchers' common language. Though Christ's figures are not weighted with the frequency of occurrence in the published docu-

ments, they are quite close to those of ours and Kraft's. This similarity suggests that the

natural language or title words are much closer to the technical terms than LCSH in the

*' His pioneering article is now easy to access through the facsimile reprint in Schultz [21].

*' Kraft [11] p. 50. *' ostecky [18] p. 303.

HITOTSUBASHI JOURNAL OF ECONOMICS

social sciences and that the utilization of the title words instead of the subject headings

like KWIC index makes it easier to get access to the technical terms.15

Table 5 shows that the descriptors assigned to one title is about 6.2 on average in case

TABLE 5. AVERAGE NuMBER OF DESCRIPTORS To ONE TITLE

Kinds of difiierent LCSH

Total number of LCSH assigned

Number of titles assigned with LCSH

Percentage to total titles in the bibliography

Average number of LCSH to one title

Kinds of different keywords

Total number of keywords in KWIC

Number of titles in the bibliography

Average number of keywords in one title

Ll rary an. mformation sc]ence

6 54

2010

ll94

66.9% 1.5

2878

1 5055

1785* 6. 2

Management sciences

63 5

1789

1 136

1.57

1511

7340

1 189 6. 3

* Number of titles indexed in KWIC index includes 616 articles in Encyclopedia of Library and Information Science (voL 1-ro). Thus total number indexed amounts to 24cn.

of KWIC index and about 1.6 in case of LCSH. Thus the title words will bring out more

access points than the usual subject heading approach.

As KWIC index requires no librarian's work to examine the content of the book, the

above comparison of natural language and LCSH might lead to a conclusion that the KWIC or mechanized information processing is superior to the subject heading approach.

But we should take the following points into consideration; the descriptiveness of the contents by the title decreases with the sophistication due to the writer's literary taste.

In this case some additional descriptors are required. This compensates the labour saving

character of KWIC index making. Kraft's experiment mentioned above says that 10.5~ titles need this kind of correction and in K. Janda's case on American Political Science

Review it is 33%･ Other experiment like Aldous & Hill's case also reports necessity of the correction of this kind, though no statistics is reported.16

Supplementary descriptors for KWIC index are required especially for the empirical

research dealing with specific time and space elements. K. Janda's work on political

science, Aldous & Hill's work on family and marriage and our work on econometrics will

show a good example on this point. In most cases, except ours, numerical figures are assigned as stopwords. The date

in the title are excluded from the KWIC index and only verbal expression of time or period

remains. However, the pronounciations of numbers and their spellings are so different

even among European languagesl7 that diversification of the same word is too wide to be

used for retrieval.

** As to the application to the natural sciences, see Ruhe [20] and Rosenberg [19]. =6 See Kraft [1l] p. 51 and Janda [10] p. 57. Also see Aldous [B-I] p. 5.

" xcept this point merger of different languages causes no serious difficulty in the case of European languages including Slavic. In case of Japanese the situation is rather different . Transcription via Roman

alphabet is not applicable for Japanese, Chinese and Korean to be treated simultaneously.

1975] ErFECTIVENESS OF KwlC INDEx AS AN INFORMATION RETRIEVAL TECHNIQUE FOR SOCIAL SCIENCEs 25

Examining these previous results, we excluded numerical figures from the stopwrds.18

In the case of econometrics the dates shown in the title are concentrated to those in twentieth

century (about 40 titles) and the rest are 1 3 titles on the nineteenth century (including two

covering eighteenth century) and two exceptional titles : one is Gossen's biography 1 8 lO-

1858 and the other is British historical analysis of 1790-1859. To obtain the information

of these 55 titles occupying 105 lines, we should scan 308 lines in the KWIC index. Simi-

lar results are obtained in the cases of management sciences and others shown in Table 6.

TABLE 6. THE ROLE or NuMERICAL FIGURES IN THE TITLE

Total lines of numerical figures in KWIC index

Figures for the years and century

i) Description of the contents

ii) Century

iii) Date of the Conference, etc.

Others (editions, etc.)

Total titles retrieved by i) & ii)

Management Sciences

303

90

78

12

212

9

Econometrics

307

1 15

6 O

1 09

1 92

55

On Saint-Snnon & Saint-Simonians

98

70

70

28

40

In the case of LCSH no exact dates are given even for the documents on history. The

analysis especially on the contemporary phenomena requires exact specification of date

and the figures in the title or in the sub-title are quite effective for this specification. But

for this purpose, the assignment of such a date in the title is not sufficient and we should

add some supplementary descriptors in figures.

As to assigning the area specffication, LCSH is superior to the direct use of the title

words, because in most cases those works written for writer's own countrymen seldom use

his country's name in the title. See Table 7.

Thus additional descriptors for area designation are required for KWIC index used

by the users aiming international comparison. Maybe for those whose primary concern

rests on his native country such additions are unnecessary. For example K. Janda assigned

AMERICA. AMERICANS, US as stopwords. His presupposition might be that articles in American Political Science Review deal mainly with American affairs.19 But when a bibliography becomes more international and extensive through inclusion of the articles

or books written by foreigners, this kind of omission of the keywords becomes obstacle

18 Our latest tiral, Iibrary and information science seen in [B-6, no. 4], is requested to make a further improvement for its data base structure includes some figures useless for retrieving due to the inclusion of

the contents of the documents like Encyclopedia of Library and Information Science. Thus the figures after V. for volume, P. for gapination, VOL. ED., NO., etc. are assigned as stopwords to be excluded from KWIC column. And figures like nineteenth century are transcripted into 19TH CENTURY. This technique is a variant of KWIC index. Similar trials are found in Matthews [17].

19 He did not make any explicit reasoning on this point. But he did not exclude UNITED STATES, U.S, and U.S.A, from keywords. We suppose that he needs UNITED, and U. to get UNlTED NATIONS and U,N. as keywords. As to the case of Aldous and Hill no information on stopwords is given as men-tioned before, this prevent us to evaluate this biliography exactly. It should be remembered for the com-piler of KWIC indexes that the exposition of the list of stopwords is necessary not only to show the compilation process but also to make a thorough retrieval.

26 HITpTSUBASHI JOURNAL OF EcoNoMlcs [February for effective retrieval.

Our tentative conclusion is that KWIC index supplemented with dates and places will

be a powerful substitute for the hand-made index like LCSH. And it may act as a more powerful retrieval technique with these supplementation and it will rarely meet incongruence

between the librarians' conceptualization and that by the social scientists.

TABLE 7. C0-0CCURRENCE OF

TITLE WORDS AND

NAMES

LCSH

OF PLACE AND

(MANAGEMENT

PERsoN AND

SCIENCES)

TIM13 IN

source: Matsui [15] p. 298.

The second point that should be taken into consideration is that, when the vocabulary

is uncontrolled, the natural language of the same meaning will be scattered throughout

the whole KWIC column. For example, VALUE THEORY-PRICE THEORY. IN-TERlNDUSTRY ANALYSIS-INPUT-OUTPUT ANALYSIS-LEONTIEF ANALY-SIS, etc. This problem will become more serious in the case of merging the different

languages like WIRTSCHAFT-OEKONOMIE-ECONOMIE, or HISTORY-HISTOIRE-ISTORIA-GESCHICHTE, etc.

But as far as European languages are concerned our experiment shows remarkable success because their differences in spelling concentrate to the ending part showing in-

flexion, etc. This, however, cannot be applicable to Romanized items in Chinese, Korean and Japanese which share common letters in part.

Needless to say, it is impossible to retrieve all the relevant documents by KWIC index

without thesaurus or controlled vocabulary lists. This defects cannot be driven away by

means of SH approach because of the lack in hierarchical categories in SH. Thus we should

make reservation for the final evaluation of classification scheme approach.20

20

the

See "Panel discussion," in Wellisch [24] and the future possibility of information science suggested in

articles of Elias [6].

1975] EFFECTIVENESS OF Kwlc INDEx As AN INFORMATION RETRIEVAL TECHNIQuE FOR SOCIAL ScrENcES 27

II. Retrospective Search and Content Anlysls through KWIC Index

As mentioned above, interdisciplinary approach may become one of the most im-portant aspects of our day and stimulate our interests in computerised indexing. And

similar approach has occurred even in the field of analysis of social and economic thoughts.

It is required to give the whole picture of one person who participated in several move-

ments in the history of thoughts or one person whose activities and thoughts created a sort

of a sect of social movements with international influence. Such a "whole prcture" rtself

changes m accordance wrth the change of "paradigms " Thus the change of analysis itself becomes an object of research. These kinds of research work requires much more

intensive bibliography than ever compiled. Recent close international contact among librarians and researchers has made it possible to compile such a bibliography in the world

wide scope. To take for example, Anthony P. Campanella's Giuseppe Garibaldi e la Tradizione

Garbaldina, una bibliografia dal 1807 a/ 1970 includes about 16,141 documents written in

54 Ianguages and dialects and Burton R. Pollin's Godwin Criticism; a synoptic bibliography

covering 1783-1966 includes about 3,374 documents in 14 Ianguages.21 The latter is com-

piled in machine readable form and provided with 1 2 computer made indexes and statistics

tables. As far as we know, this may be the first trial of computerised bibliography designed

for retrospective search for social sciences. But unfortunately this work sticks too much

to the traditional form of a bibliography and does not realize the potential power of the

machine readable form. Even in the field of humanities the computer processing of non-

numerical information has come to be a strong weapon to compile a concordance or to identify the authorship of the anonymous or pseudonymous texts through quantification

of the contents of the texts. Such a technique of analysis of non-numerical information

may be applicable for the bibliography in machine readable form like Godwin Criticism

equipped with short abstracts and many descriptors evaluating the degree of relevance to

Godwin and / or his works. Abbreviations used for descriptor are A; article on Godwin,

R; review of a book by Godwin, P; passage on Godwin, C; short comments, interspersed,

M; mention of Godwin; B ; book solely on Godwin, N; necrology of Godwin, Q; quoted

material from Godwin, W; work dealing in part with Godwin and combinations of them.

The chronological lists of all entries is a mere listing of item number accompanied

with these abbreviations. But a cross-tabulation with languages used or place of publi-

cation could have shown Godwin's international influence quantitatively. And chrono-

logical listing of quotations showing from what material of Godwin one had quoted would

have shown what aspects of Godwin had attracted the public attention. This may be a

variant of the citation index often used in natural sciences.

Pollin's treatment lacks a taxonomic point of view. Although his first version before

publication classified all the material into three groups i) those written in the days when

Gowdin was working actively, ii) in Victorian period and iii) in modern times, he suppressed

all these distinctions into two sections before and after 1 836 because of the smallness of the

'* See Campaneua [B-21 and Pollin [B-5].

28 mTOTSUBASHI JOURNAL OF ECoNoMlcs [February amounts published in Victorian period. But the waves of the number of publication are

the reflection of the fashion of thoughts of the day. Relevant classification will reveal the

change of thoughts more clearly. KWIC index applied to the titles of each period might

suggest the difference of thoughts more ,clearly. For the selection of the title words are

governed by the frame of reference of the original writer. And this frame of reference is

difficult to escape the fashion of the day. Thus the analysis of the difference of the distri-

bution of keywords will supply the quantitative basis for this comparison.22

One might say that these suggestions were beyond the scope of a bibliography. But

such an intensive bibliography as Pollin's one itself provides one interpretation of Godwin's

thought and influence and implications in Pollin's bibliography should be fully visualized

through quantification of the bibliographical contents and abstracts.

We will show this kind of quantification through our experiments on Saint-Simon and

Saint-Simonians' works. Before going further it may be allowed to make some digression

on late Professor Juro Tedzuka whose endeavours, as a result, had provided the basis for

our experiments. Although he is known as one of the most excellent pioneers of mathe-

matical economics in Japan, his concern was not restricted to the mere refinement of

mathematical treatment of economic models but extended to the philosophical implications

underlying the economic theories and the possibility of the social reform from the side of

libertarian favouring P.-J. Proudhon.23 And he spent his stay in Paris from 1921 to 1926

in collecting books and brochures and copying unobtainable ones at Biblioth~que Nationale

for this purpose. His collection covers from Pierre Bayle. Castel de Saint-Pierre, Ency-

clopedistes to contemporary radicals Andre Lorulot and Max Nettlau, especially strong

in the nineteenth century visionaries like Saint-Simon, Fourier. Cabet, Considerant and

their followers. After his death his whole collection was donated to Otaru University of

Commerce by the courtesy of Miyakichi Itaya. It is on this collection that our experiment on Saint-Simon and Saint-Simonians'

works was undertaken. What he left was not mere books but he also left us one disciple,

Mr. Kiyonari Kidahashi who compiled the first catalogue of Tedzuka collection and de-

signed a hand-made KWOC (Key-Word-Out-of-Context) Iike index for the section Philo-

22 One my easily find this way of investigation as a variant of a quantitative symbol analysis or, more

generally, a content analysis. For example, see Richard L. Merritt. Symbols of American Community,

1735-1775. New Haven & London, Yale University Press, 1966. If we extend our scope beyond the bibliography making, much suggestions will be obtained through

the trials in the field of the computerized information processing found in the journals like Computers

and the Hurnanities edited by Queens College since 1 967.

'* Tamotsu Matsuura describes him as Toshiro Tezuka not as Juro Tedzuka in his "Marginalism in Japan" in The Marginal Revolution in Economics; interpretation and evaluation edited by R.D.C. Coats

& C.D.W. Goodwin, Duke University Press, 1973. But we followed Tedzuka's own alphabetization used in his articles like a contribution to W.L. Valk, The Principles of Wages appeared as a Bibliography

annexed by J.Tedzuka. As to his life and works, see an annonimous article, "Recherche d'~conomie politique du professeur

Tedzuka et son m6rite," in Bibliographie des Id~es sociales et economiques en France,' Catalogue de [a

bib!iothique Tedzuka de l'universit~ d'Otaru, Redacteur Taro Sakata, Tokyo, 1966.

1975] EFFECnVENESS OF Kwlc INDEx As AN INFORMATION RETRIEVAL TEcHNIQul3 FOR SOCIAL saENCEs 29

sophie et science naturelle in late forties.24 His trial found no supporter or collaborator

and was isolated until we found a precursor of H.P. Luhn in him. And after twenty years interval this small brochure of about one hundred pages drove us to extend his intention through computerisation.

The importance of Saint-Simon and Saint-Simonians' role in the history of social and economic thought requires no additional explanation. Our da{a base is composed of the

bibliographical description of books and pamphlets written by and on Saint-Simon and

Saint-Simonians excluding articles in periodicals. The data base is subdivided into two

files ; one is for the books written by Saint-Siomn and Saint-Simonians themselves (here-

after called BY-SS data base) and the other is for the works written about Saint-Simon

and Saint-Simonians' works and their activities (hereafter called QN-SS data base.) BY-SS

data base is not complete and we are preparing the second version through international

cooperations. Thus the results obtained are rather tentative. ON-SS data base is based

on three major information sources; i) 70 items through direct access to the books in Ted-

zuka Collection and 6 items in L60n Centnell Collection now in Otaru University of Commerce as Ohnishi-Tedzuka Memorial Library,25 ii) 1 3 1 items depending on the des-

cription of the secondary information source, Walch's Bibliographie du Saint-simonisme,

and lii) 30 itmes from various information sources including our private collections,

Indexes obtained from these data bases are i) KWIC index, ii) chronological index,

iii) title index and iv) author, index. Other information produced from these data bases

is vi) matrix of publication place and date, vii) matrix of language and date and vlii) various

statistics based on these indexes, etc.

Table 8 is the distribution of publication date obtained from BY-SS ON-SS chrono-

logical indexes. This clearly shows the interaction of BY-SS and ON-SS. The upheaval

of the sect of Saint-Simonians brought out a burst of publication in 1 830s and their revivals

are found in BY-SS data three times ; at the end of the nineteenth century, after World

War I and in 1960s. In most cases the waves of ON-SS data follows BY-SS data. First

" This is an index of 888 headings to 1993 titles for the Section; Philosophie et science naturelle of Tedzuka's

collection. An facsimile example is shown below. See Bibliographie de feu prof. Juro Tedzuka.' Section Phi!osophie. Tabie analytique, [r6digie par Klyonari Kidahashi], x, Ioo p. (copie carbon~e), [n,d., otaru].

Binet Altred (1897-1911) . ' --. ･Naptin.R. Paris,192k.pl)80.

Biologie --

~nl~:,ale./ El6nente de --. -Raba'ta,!.l~. 1920.p908.

--･/ Leg li!uiteg de la --.~~･asset.J.191?･p555. Biol08ique en Eupope./ Le ~ouveGent -.-Bohne.G.pl55. --･/ Lg fcrce et le dpoit le pr~tendu dpoit -.

-Antheny.R. Paris,1917･ p5Ol. -- et sociales./ Les wathel!'etiques daas le5 scienoes --. -Voltepra.V. l,06. pl725.

Biologiste sup l'ebJeet, Ies !QI~thodes et les liuites de la psychologie./Rerexions d'un --.-Antheny.R. p57-

Blonde l.,caupice ( 1861-19~9) -

--･/ L'intelli8enee dtap.~s --. ~lapitain.J. pl;9-

25 This collection was donnated by the alumni to commemorate the sixtieth anniversary of this university.

As to the detail of the collection, see Matsuda [13].

30 HITOTSUBASHI JOURNAL OF ECoNoMlcs

TABLE 8. NuMBER OF PUBLICATIONS CLASSIFIED BY DATE

[February

* This figure includes many brochures whose publication dates are assigned to our estimation mainly based on Fournel's Bibliographie saint-sirnonienne, de 1802 du 31 dicembre 1832. 1833.

upheavals in 1830-50 is the result of the controversies and interests of the contemporaries

of Saint-Simonians. The second upheaval at the end of nineteenth century was brought out by the publication of (Tuvres de Saint-Simon & d'Enfantin which made possible to utilize

many unpublished documents. It should be noticed that the KWIC index shows the increase of SOCIALISME, FEMlNISME, POSITIVISME. This may be the change of frame of reference of this period. The second increase in 1920s may relate to Neo-Saint-

simonism and the Russian Revolution. Increase of monographs in English in 1940s may

reflect the totalitarian political movements of the day. These changes of interests of the

time are seen in the KWIC index, and the difference of frame of references between Saint-

Simonians and researchers is also clearly observed in the KWIC index. For example,

Saint-Simonians preferred the words, ORGANISATION, ASSOCIATION, RELIGION etc, to SOCIALISME, UTOPIE etc.26

Further application of content analysis requires much more elaborate data base which

at least includes additional descriptors or abstracts. Another limitation of this data base

is its exclusion of other synchronic social reformers' works, which are necessary to follow

their mutual influences. Tedzuka Collection and Ohnishi-Tedzuka Memorial Library are qualified enough to provide materials to compile a data base suitable for this purpose. But

as our initial concern was limitted to estimate the effectiveness of the automatised indexing

and abstracting processes, such kind of elaboration of data base is postponed to the next

stage of our experiment. Some words should be added as to the effectiveness of a retrieval technique of this

kind. In the previous section we made a reservation for the final evaluation of classifi-

cation scheme approach in case of interdisciplinary approach. But in the case of

" See also Matsuda [13] and Sato's introduction to Matsuda [B-6, no. 3].

1975] EFFECTIVENESS OF KWlc INDEx As AN INFORMATION RETRIEVAL TECHNIQUE FOR SOCIAL sclENcEs 3 1

retrospective search the superiority of mechanized information retrieval over classification

scheme approach is quite clear. Evolution or expansion of classification scheme is itself

a result of the division of research disciplines and change of the frames of reference.

This means the applicability of older classification scheme to older documents which will

be retrieved in retrospective search. Unfortunately change of frames of reference differs

from topics to topics. Thus the mechanical conversion or application of the classification

scheme has no practical meaning.

III. AssOciation of Keywords and SDI System

Along with KWIC index system, H.P. Luhn had developed the idea of SDI (Selective

Dissimination of Information) System where newly arrived data base will be screened at

once through a compound list of the descriptors or keywords which each user had pres-

cribed in order to avoid repeated scanning of the same machine readable data many times

for each user. The practical applicability of SDI to social sciences might be severely

limited because of the lack in massive machine readable data. But even if such data were

available, there would remain another factor which will suppress the effectiveness of SDI

for the social sciences. For, exact meaning of the categories used in the social sciences

often depends on the context and exact specification of descriptors is quite difficult.

The same factor will, in turn, favour the KWIC index. The effectiveness of KWIC index depends on the degree of the context-dependency of the category shown by the key-

word. Exact meaning of a keyword will be brought out by the association with other

words surrounding the keyword. Thus the effectiveness of KWIC index and similar techniques cannot be determined uniquely. It depends on the user's insight. This fact

is applied to another aspect of a KWIC index retrieval process. The effectiveness will be

increased with the association of keywords27 through glancing the column of keywords.

Some might say that this kind of arbitrality is a defect of the system. But this arbitra-

lity should be regarded as the flexibility of the system which can create the room for learning

process in retrieving process in action. For the rigid SDI system provides new informa-

tion aspect only through assigning new descriptors and so learning process at the retrieving

process occurs within a limit of user's memory and documents retrieved previously. On

the other hand, glancing the column of keywords will supply a new frame of association

of categories which might be overlooked by the researcher in the case of SDI system.

Our hypothesis underlying this way of thinking is that it is impossible for the searcher of

information to grasp the whole idea of what he wants to retrieve. This might sound ridiculous but is much closer to the reality. Thus the effectiveness of any retrieval technique

still rests on the skill or insight of the user.

Another limitation to the information retrieval system, however, comes from the extent

of the exhaustiveness of the documents to be retrieved and the degree of additiQns of descriptors or abstracts which are utilized for retrieval. Like production of goods, no

input creates no output but inappropriate input creates no output either. Similarly no

" The association here defined is not the same as to the association found in Lauren B. Doyle's "Index-ing and abstracting by association" in Elias [61 but may correspond to the abduction process of perception.

32 HITOTSUBASHI JOURNAL OF ECONOMICS

AppENDIX I . MAlN

[February

FEATURES OF THE

Reference no.

Organizer

Publication

Subject field

Coverage of publication date

Coverage of publication form; A, articles. B, books.

Source materials ; D, Direct access or primary

sources. S, secondary

sources

Selection criteria; M, me-

chanical, S, selective

Treatment of languages other than English

Items covered

Data base compiler

Notes

3 CB-5]

Burton R. Pollin

l 967

On William Godwin

1783-1966

A,B

D,S

S

Original language in Ro-

man alphabet trans-cription with English

translation

3,379

B.R. Pollin with inter-

national corporation

1975] EFrECTIVENESS OF KWlc INDEx As AN INFORMATION RETRIEVAL TECHNIQUE FOR SOCIAL saENcEs

BIBLIOGRAPHIES DISCUSSED IN THE TEXT

33

34 ruTOTSUBASHI JOURNAL OF ECONOMICS [February

Reference no.

Master list

Indexes Name approach Author

Persons

Title

Titles

mentioned

cited

Periodical title

Subject approach Natural language

Descri ptors

C]assified scheme

Evaluation approach

Chronological approach

Language a p proach

Number Number

of indexes of statistics

Computational work Number of data cards Format of masterlist Pages (catalogue, illustra-

tions)

Computing facility

Computer Programming Programmer

language

1

**Author alphabetized bibliogra phy (listing by alphabetiz-ed authors)

*Author-cross reference (author and coauthor)

+

**Keyword listing (i.e. KWIC index)

3 o

n,a. B-5, 90 Iines/1 column (225, 1 2)

Northwestern Univ. Computing Center

IBM 709 MAP, FORTRAN James S. Aagaard

2

**Complete reference list (listing by alphabetiz-ed authors)

*Author list (author and coauthor)

(*)Periodicals list

(not index)

**KWIC index

Subject index (i,e. clas-sified index)

- Only statistics)

- Only statistics)

5 5

50,000 above A-4, I 07 Iines/2 colurnns (225, 1 2)

Univ, of Minnesota Data Processing Cen-ter

IBM 1401 n.a. Ralph J. Willard, Ernst Duffing

3

**Synoptic bibliography (classified by publish-ing dates before and after 1837, books or periodicals)

*Authors of books and articles

*Persons mentioned (selective)

*Books mentioned (selective)

*(Combined with the synoptic bib. Iisted by alphabetized periodi-cals)

*(Publication form)

**(Abbreviations & short abstracts)

*Chronological listing of all entries, Writings

of Godwin *Languages of original

sources

11 3

20,000 B-5, 58 Iines/1 column (659, 46)

New York Univ. The Institute for Com-puter Research in the Humanities

IBM 360-30 PL/1, FORTRAN George W. Logmann

Notes: *

**

+

Index indicating the

Index combined with Not tried.

Not applicable.

item number. short bibliographical description.

1 975] EFFECTIVENESS OF KWJC INDEx AS AN INFORMATION RETRJEVAL TECHNIQUE roR SOCIAL SCIENCES 35

36 HITOTSUBASHI JOURNAL OF ECONOMICS [February

information will be obtained by inappropriate data base. Thus the real problem is how

to compile data base for a bibliography or catalogue. '

Apart from this problem of adequacy, the exhaustiveness or thoroughness of coverage

requires a close contact with various institutions in international scope in the case of the

social science bibliography which, more or less, contains some historical and international

aspects. This requirement will be fulfilled by the process similar to compiling a subject

union catalogue. Unlike the national or aggregated union catalogue, subject union cata-

logue requires selection from the whole holdings of institutions participating in the union

catalogue making. In this case, not a card catalogue but a check list equipped with many

indexes will decrease the burden of librarians to corporate with the program, because a

book form catalogue is easy to scan. Thus, not a mere listing of the items to be searched

but a bibliography fully equipped with indexes like a final draft is necessary as a check list

for the holding survey. For this purpose, the machine readable cataloguing technique is the most powerful weapon for editing. For, once several indexes are prepared, often

changes of the item numbers causes serious difficulty in the case of hand made catalogue.

So in most cases tentative drafts are mere lisitng of whole items to avoid the overall re-

numbering which will be caused by the adjustment of the item numbers for the new additions or deleting. But, by virtue of the computer processing, this kind of adjustment

comes to be a fairly easy task and so the circulation of the drafts of bibliography is possible

for any version of the drafts. While it is not quite easy for computer processing to prepare

the drafts which can be circulated several times with ease, recent development of COM

(computer output microfilm) device provides a good means of producing a circulating version.

Thus our final remark is that the future of the computerised bibliography making depends on the piling up of the trials of compilation of such bibliography in practice. We

hope this short essay will be a stimulus for the future trials.

AppENDlx 2. DATA BASE AND SOME TECHNICAL NoTES

The data base for [B-6] was first restricted by the compiler of PL / I of FACOM 230-60.

The area of characters is within 256 Ietters, and so the title of the document is restricted

within 256 Ietters including blanks between words. The input format is fixes field and one

record is 1,000 Ietters in case of [B-6, no. 1-no. 3]. Extension to 1,400 Ietters is executed

for [no. 4] to utilize note area.

The master file format is shown in the Figure 2, below. As this is a final print out format,

the end of the line is adjusted by the hair space technique. Fig. 3 shows the example from

[B-6, no 4] which doubly prints out the titles having articles at the top as omitting article

case and including article case. As to the detail of the programming techniques, see Matsui [16].

1975] EFFECTIVENESS OF Kwlc IN)Ex AS AN INFORMATION RETRJEVAL TECHNIQUE FOR SOCIAL saENCEs

FIG. 2. DATA BASE

Management Sciences (not yet published. Ref. no. 5 in Appendix 1)

TARSK I ･ALFRED I N1' RODVCT I ON TO LOG I C ANO TO THE ME THODOLOGY OF DEDUCTIVE SCIENCES･-

1941 NEW VORK LCSHI MATHEMATICS--PHYLOSOPHY LCSH2 ARITHMETIC--FOLINDATIONS

[B-6, no. 1]

2,2･ J

CHXA'~A ' XAIUSH r . ,DYNAr i c ANALYS I s OF NOGYO No DOTAI BUNSFXl

AGR I CUL TURF ) ･ ' 1g'Jl, Toxvc:Josul sHc8c, (F-FFTR 27- ,- ,~97 AEV.8Y NAkAPURA,JAF,ES [,)

2,,~ F

ALLA I s ,~dAtlR t CF ,

FONDEPFNTS D'uNE THEORIE PCSITIVF OF~ CHCIX ceppCRTANT Uis RlsQuF FT CRtTIOUE DES ~C~TULATS FT Axlr,$Es OE L'FCCLF A~EAiCAINE,'

i95, ,5P PARls: iMPRIHERIE hJATTONALF , (F-FFTR 27- ･- 49G REV,8Y su~pFs,pATRICx.)

2525 E L l. l. ~27 HEADE , J. E .

TPAOE ANO t,ELFARE.(~ITH t~ATHEP'ATICAL suPPLFrENT)' 19,5 ~ltP LO?~IDCN:OXFORO U~JIVFnsl7Y PREss, (E-,,FTR 27- ,- 5oo REV.BY HARBFRGFR,AANCtD c.)

37

[B-6, no. 4]

O059 J cL: OOIC HOxxA:OO･ MUSASHI J05HI TANKr DAIGA,(u. TOSHOXAN.

TOsroxAN JCHOGAXU 5HIRYO ,10,(uRO,(u. --1572 NEN 5 GATSU c;ENzA[--.-

IS,72 S,LPPOR0: TH8 L!BRARY･ 77P.

LOC･: OTARU U O1 ,1USASH: Jc

-- 02 ENCYCLOPEOIAS. DICTIONARrES, GLOSSARIESt HANDBOOKS ANO ,,ANUALS --

( EbC(cLOPeO I AS )

oo.,O E cL: 0020 O il- 3- l.,3 XENT. ALLE,J (EO) LANCOUP. HAROLO (EO) NASPI. ,,･ Z･ (AS EO)

EhCYCLOf.EOIA OF L!8PARY ANO rNFORPIATION SCIENCE. VOL 1: A TO AS50CIAC.-

1968 ~C,d YORK: MARCEL OEKKER･ 12. 676P-

LCSHI LI3RAPY SclEhCE--OlcTIONApfEs LCSH2 I,fORMATfON 5CIENCE--OICT:ONARrES LOC.: OTARU U･C,1 HU(CTR) HU(ENe) HU(EOC)

HUEcsAP) XITAP11 1 ･T 5AP･MED･C. HU(AGR) HOxXAI 9J HOxUSE: GV SAPPORO U･ MUSAsHf JC SAPPORC JC HOx.Pf?E.La UCIXANAGAH 5B/BgLZER

(O1) A88~vIATIObS (EL15 v.1 P.]-12) (STERNaERG, vfRGlwrA)

(02) A8STRACT CLA5SIFICATiON (ELIS v., P.i2*:6) (OAILY. JAy E.)

(03) A85TRACT5 AP,O ABSTRACT:NG (EL!5 v.1 p･16-!8) (8ERNreR. CHARLES L･)

(04t) ACAOOllC STATUS OF LIBRAA:A,6 ANO INFORHAT:ON sr_~SNTI~TS (5LES V.: F･:･*･tS)

(XHURSHtO. AN!S *)

［FebmaryHITOTSUBASHI　JOURNAL　OF　ECONOMI（S

Q～一〇

拷自0

5δO

のOOO

タ必0

7ひ唱O

うあ4一

〇のOO

O鳴OO

，一

〇一

η」凹0

3卜〇一

〇〇

40

唱〇一〇

肖偶『Q

閃OOO

苗騎C一

〇卜曽O

遷O

笥QりO

ヨトO

噂030

3、bO

協O頃一

瞬，φO

O講O

q湾O

（、V（O切V　．J　Oり虚㌧一悼；　，艶翼⊃《ω

．．。bと一9〕≧Z⊃．ζ6トまくU

　　　聴遇‘↑ノ戸『．G，↓5’OUJ　．甕パ》｝q

　　　口8（峯　OJごプσ届」　。垂

（、）　　　　〕圏¢⊃望　《0〕一に」　．α田JJ婦工

（うXOり）5≧《J蒼髭、O　ヂお2六『

　　　　　　　　　　　，一つく諏　　だ謹］厨く‘

（う）　　　　　　（9）　ダ（y　．暖り，一」」Q」

　　　　　（8）　工旧こ」《需　

D畢星一冥

　　　　　　　　　　　　　9く　》ε　　・トト矩∩『

（う〉　　　　　　　　　　　　Z〕一」雛つ　『〈周く∪

　　　　　　．〉　裏』罵り　ひ』里場」「望

　　eO一一受　ω「昌リムトO一，、Q3H鳴

（う）　　　　　　　　翼ンぜ

ハつV　ω3《卜嚢　OJ善U一

（「）　　　　　　　　　　　XJ里畠解Hヱ

（うV　　　　『萎＞〕　髪－Oω

　　　　　　　　　　　　　　α旧卜困

．，．O一ト《ζ　U「d里一，O閂超剛め

（「V　葦…一

　　　　　　　　．」　一〕，涙》一、■

　　　　　　oく「嘱（8ア』鯉戸

（、〉　　　　　　　．（　溜乙一Z旧甲幽一く∪

　　　　　　　　　　　　馨イ回一氏

　　　　　　　　　　ZM気丸〉

つO　　の四配の工」五　　X⊃《

2への

　　　　　　・の一¢《“

　　　　　ど）曽田釜

　　　　　．垂

　　　　　訂薗の多と

　　　　　　，望」モ

　　　　　　．η達己

　　一qUqユーミ　　f叫U

　　　　　　。の一ユ《儀

　　　　　．舞士』くリ

　　　　ゾ昌器8り

q6臣ωり　・｝をQ

　　．一U，一一一工，霞O

　　　　　　のの■ム丁

　　　　　ひα回”望目

　　炉］JJ、駒、乞一《一〇

辺一一〇］O

一雛。記、蟹糞．襲藥韓什欝樋㌶い離継襲藁　』

　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　8．］OOO　ご　　．讐Uの］二工Qコ①　鴇蟹Qろと一X」」

　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　昌．Y8η　く　（〉、hqn　い，」」。一

　…

憾駈．躍羅薗劇鞭灘鞭糠繹灘嘘．煙藻戴纏蝋騒難鍵難㎜

6．

一．ωωJJ一（雀］＞

ー（卜．2

．，望卜O卜

，oqU三q《）

配U≧Zコ） 　　

　UOロX翼く一〇氏Uロ〕’ト］］」くZ弓一トゴ2国8田一〇NJ田のく」UQ馨四ト藍潭δ」窪9、・

　　

　一・瑠肛一つ肛トリOのUJUO望〕＞9丸．没コ曇塁⊃］」一切U配》一」のUOの噸x翌Oの」』し〕量「

　　

・〕」く70一ト《7ω「ごUエ一〇一」m岡m（」可Oω望桝起ら出の国O↑z田．〕ト¢くし8⊃Oの］⊃8，2、h《U

　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　置（33“　↑ω、Ioo一《しo　U⊇G〉　　．ooヨ曽工トO「呂一の

　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　重．塗ト響　》護39一」

　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　ロ．QZ⊃ト」く3配U〉uうγ｝二〇一」の一皿　α器

　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　3．O〕一傑《促口【，一　ヤO」

　　

（、‘巽αOO円JJくよO一トく謹UトZ一2くエ一Jく］コZQ翼き∪）　。のとQOOD〕釜卯」授ア9、」

－・勇崔≡u崔雪．藁馨，。の§《．嚢5一建－コくヒ．謹図きトの一＝昌川蝶一

6蚤z冨ゴ品．5髪∪呂uき〕u＝き．匂。，3．。。”琶さ雪6撃9G三bと凄

］舅翌］」歴配（詔コ巳一配《ユエヒヱ　・（9＝2（りZ一］U剛＞工宙しりト弩岳一，一り【，お浸、δの＝Z⊃

1・OU・創　・もo留のく象．．窪一ト　．塗．O一トく【）Mぼ矩凄　．〕Oく¢トとOQOε一暴Q」）ムト・6

ぎ髪目ζ注9（注Q里ト三里一呂望《鴇〕8ごヨ］ヨ琵撫召・≧Qど．

　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　曜・OZ一ト一匡3　9べ　Q客一qqU¢　Z剛　］いα8U

　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　】（3一」　卜Uうーの一くoり　U8）　　．M｝導　ωのの差

　　

認一5り。窪〕》田整。5Z。」も。］コ。Z。ニョ・乙に拐製8萄彗・…u乙器》ご

×m【（口Z一　国■』し【』r

　　　　　　　　　　　　　　　　　（ON（一　ト）

　　　　　　　　　の一匡（こ　“〇　　　　一ρの一

　　　■・Z』一ZO三のー一三（㎝四〇］」」O㌧

岩9zoΣ誘・一三（の《zo＝《）o》三

　　　　　　　　　　　　　　　　　　　Dマ一一¢〕一Z匡（0匙

＝On

　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　（Oひ∩一　一》

　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　の一こ《α　ユOO〔　　40伊一

　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　．。O（O「」．一1．【，　肛（ユ

。N田エ）⇒』・£，・「・」⊃3一〇〇」C一）Oφ一のUエ一Z＞の《」［一］トエO）］トのつO⊃《

●一Z髪き一u。〕3E」ま弦Z豊（ゴ2≡。三切－…急

95三㌢三話二9く」ごε二ま。ま置」．］・≦）。φu3≡ざユ

《一　　”一一《σoの一⊃巳二」oユ』oの一〇⊃トu．の一」oエ⊃エの国oヒo肛o］」

　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　．一1．「．一Z《一一の⊃「IO《O一し

　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　匹自O∩

　　　　　　内馬内一　の

．φ．O畠

のU『暗

笥鴇」

論UJ

もう

」

　　9』H塁』

　工）⊃角や尉（財」『

　　母h〕幽』辱《UJ

　讐一菱《煽J

　　　　　　　ωJ

　　　　　　　UJ

　　　　　い3（」

　　　肛崔」

　　冒（コ箋」

　　緊と、霊J

》5ト落o二く」

　　　　　　　く」

　　　　　　　（」

鯉鯉菱卜一ト・トトト・

里卜

（

O、一ー』一

“轟“一

くひ一

コ噂“一

〇う縛も一

ρ】瓜一

頃n“一

N矯pい一

噸｝曵

“等い一

hOひ一

〇、ホ一

ぽ5一

蟄一“一

昌論“一

鴇Oぴ一

ひ喝ホ一

n印δ一

鐸い一

一▼酬一

幹℃〇一

Nい♂一

｝も6一

笛コ一一

〇ヨい一

〇〇あ一

　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　（4ひ0〔　卜｝

　　　　　　　　　　　　　　　　　　　　　　　　　　　　　9の一αく氏　吐On　　　一nO一

　　　　　　　　　　　　　　　　　　　　　　　　　　〇．（9国国OJO，　つO　のト一（ぼトく］V

あ（一鴇（ΣZ。α語U」垂α監9畠一三乳（」切Zさ畠一≡）α（Z9

（」　のz（。　霊岩一uりz・）ト語話どσ一翫．u＝8語＝ムヒo匡o

U」雪の㌢9窃u呂9出J2一トZ一霊呈）2①山5（UのZO色山肛

　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　。御荏7Uエ．」四Zα⊃O」

　　　　　　卜試の　　　　　　　　　　　　　　　　　　　　　　　　　　」

　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　（6，0n↑）

　　　　　　　　　　　　　　　　　　　　　　　　　　　　　．uワ瞬鑑（巳　α一n　　　　h6の一

　　　　　　　　　　　　　　　　　　　一甲ZO「ト（）『JO、⊃氏　田匡U「XUα色．一乙『（Σ　（」

岩U巴」（）（」㏄孟面J」口の牙ア（Uα）（エ⊃Oに一」〕OZ≡国工）⊃O

　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　。．Σ．「一にZ〕＝，」UZα⊃O巳

　　　　　　＄oU　　　　　　　　　　　　　　　　　　　　　　」

　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　（eぜひつ∩　↑）

　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　，の㎝『（氏　　氏O〔一　　　nn⑩一

　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　－唖Nnの一

一匡〇五田）四〇　一n⊃（NOの一ω06ZZ巴ZO三のート～ぞの巴エユq鷹OOロ」O一の

　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　・。Σ。「　…Zロエ．」UZ匡⊃OL

　　　　　　O諸。り　　　　　　　　　　　　　　　　　　　　　　　　　　」

OOト∩

ひmρn

軋，ゆn

【m

．O信

、℃－色

冒．o自

づ－巴

38

1975] EFFECTIVENESS OF KWIC INDEX AS AN INFORMATION Rl3TRIEVAL TECHNIQUE FOR SOCIAL SCIl3NCES 39

REFERENCES

I. Bibliography cited.

[B-I] Joan Aldous & Reuben Hill, International Bibilography of Research in Marriage

and the Family, 1900-1964. Minneapolis, University of Minnesota, 1967. vi, 508 p.

[B-2] Anthony P. Campanella, Giuseppe Garibaldi e la Tradizione Garibaldina, una biblio-

grafia dal 1807 al 1970. Genevra, Comitato Dell'Istituto Internationale di Studi

Garibaldini. 1971. 2 vols.

[B-3] Malcolm H. Gotterer, KWIC Index, A Bibliography for Computer Management. Princeton / Brandon, System Press, 1970, viii, 152 p.

[B-4] Kenneth Janda (ed.), Cummulative Index to the American Political Science Review,

volumes 1-57; 1906-1963. Evanston, Northwestern University Press, 1964, xxili,

225 p.

[B-5] Burton R. Pollin, Godwin Criticism, A Synoptic Bibliography. Toronto, University

of Toronto Press, 1967. xlvi, 659 p.

[B-6] Yoshiro Matsuda (ed.); KWIC Index Series for Social Sciences.

No. 1. Bibliography, Works on Econometrics reviewed in Econometrica, 1951-

1957 (Introduction by Yoshiro Matsuda & Sachiko Matsui), Otaru, 1973. vii, 24, 145 p.

No. 2. Bibliography,' Works by Saint-Simon and Saint-Simonians. (Introduction by Akira Hirota and others) (to be published)

No. 3. Bibliography, Works on Saint-Simon and Saint-Simonians. 1831-1971 (Introduction by Shigeyuki Sato & Sachiko Matsui) Otaru, 1973. viii, 24, 49 p.

No. 4. Bibliography, Works on Library and Information Science, based on the holdings of the libraries in Hokkaido region. Compiled by RlNET (Regional Information Network) Society with the joint auspices of Hokkaido Librarians'

Study Circle. Otaru, 1974. 2 vols.

II. Books and articles cited.

[ I J Wm. M. Adams, "Relationship of keyword in titles reference cited." Amencan Documentation. Vol. 18, no. 1967. pp. 26-32.

[ 2 J H. Borko, "Me~suring the reliability of subject classification by men and machines."

American Documentation. Vol. 15, no. 4. 1964. pp. 268-73.

[ 3 1 John M. Christ; Concepts and Subject Headings; their Relation in Information Ret-

rieval and Library Science. Metuchen, N.J., The Scarecrow Press, Inc., 1972. 174 p.

[ 4 J J.E. Daily, "Many changes. No alteration; an analysis of Library of Congress Sub-

Ject Headings 7th edition." Library Journal, Vol. 42, Nov. 1967. p. 3962.

[ 5 J Richard De Gennaro, "Harvard University's Widener Library Shelflist Conversion

and Publication Program. College and Research Libraries. Vol. 31, no. 5. 1970.

pp. 318-31.

[ 6 J Arthur W. Elias (ed.), Key Papers in Information Science. American Society for

Information Science. Washington D.C., 1971. 223 p. [ 7 J M. Fischer, "The KWIC index concept; a retrospective view," American Documen-

tatron Vol 17 no 2 1966 pp. 57-70.

40

[8]

[9]

[lO]

[1 Il

[12]

[13]

[ 1 4]

[15]

[ 1 6]

[ 1 7]

[18]

[ 1 9]

[20]

[2 1 J

[22]

[23]

[24]

HITOTSUBASHI JOURNAL OF ECONOMICS

H.H. Goom, "A computerbased current awareness system producing both SDI output and conventional abstracts bulletin." Aslib Proceedings. Vol. 26, no. 3.

1974. PP･ 98 108 G. Jahoda & L.M. Stursa, "A comparison of a keyword from title indexes with a

smgle access pomt per document alphabetical md x." American Documentation. Vol. 20, no. 4. pp. 377-80. Kenneth Janda, Information Retrieval; application to political science, New York,

Bobbs-Merill Co. Inc., 1968. xxiil, 230 p. D.H. Kraft, "A comparison of Keyword-in-context (KWIC) indexing of titles with

a subject heading classification system " Amencan Documentatron Vol 15 no 1. 1964. p. 50.

Yoshiro Matsuda, "No Deus ex Machina will appear in our library modernization."

Keizai Shiryo Kenyku, (in Japanese), no. 4. 1971. pp. 1-23.

ditto, "Retrospective search through KWIC index; Bibllography and library for

social sciences." Shogaku Tokyu (Economic Review, in Japanese), New series vol.

24, no. 2. 1973. pp. 1-32. ditto, "Present state of documentation in Japanese social sciences." Joho Kanri

(Information & Documentation, in Japanese). Vol, 17, no. 6. 1974. Sachiko Matsui, "Information retrieval for interdisciplinary science through subject

headings of Library of Congress. Case study of management sciences." Shogaku

Tokyu (Economic Review, in Japanese), New series vol. 23. no. 2 / 3･ 1971, pp. 287-310. ditto, "Programming techniques for machine readable bibliography. Review and

case study in the field of social sciences." Shogaku Tokyu. (Economic Review, in

Japanese) New senes vol 24 no 2 1974. pp. 33-70. F.W. Matthews & A.D. Shillingford, "Variations on KWIC." Ashb Proceedmgs Vol. 25, no. 4. 1973, pp. 140-152. V. Mostecky, "Study of the See-Also reference structure in relation to the subject

of mternational law " Amencan Documentatron Vol 7 1956. p. 303. K.C. Rosenberg & C.L.M. Bolcher, "A comparison of the relevance of KWIC vs.

descnptor mdexmg terms " Amencan Documentatron Vol. 19, no. l. 1968.'pp.

27-29.

M.J. Ruhe, "Chemical documents and their titles; human concept indexing vs. KWIC

machine indexing. American Documentation. Vol. 15, no. 2. pp. 136-149. Claire K. Schultz (ed.), H.P. Luhn; Pioneer of Information Science; Selected works.

New York, Spartan Book~, 1968. U.S. Library of Congress, Subject Cataloging Division, Processing Department. Subject Headings used in the Dictionary Catalog of the Library of Congress. (ed, by

M.V. Quattelbaum) 7th ed. Washington D.C., Government Printing Office, 1966.

1432 p. B.C. Vickery, On Retrieval System Theory. London, Butterworths, 1961. Hans (Hanan) Wellisch & Thomas D. Wilson, Subject Retrieval in the Seventies, new

directions, Proceedings of an international symposium held at the Center of Adult Educa-

tion University of Maryland College Park, May 14 to 15, 1971. Westport, Greenwood

Publishing Co., 1972. (Contributions in Librarianship and Information Science. No. 3.)