Class index: A new approach to information retrieval

INFORMATION RETRIEVAL

A post-coordinated indexine system withUDenumber as descriptors is proposed. The need for numerous cross references to inter-relate multiple descriptors for denoting a subject is eliminated by the use of class numbers and their coordination at the time of retrieval. Moreover this can get over the language barrier in indexing as the UDe numbers would pro- vide an international language.

Introduction

Implications of this system are simple.

In uniterm index. terms or keywords are matChed. while subject headings or subjects are coordinated in Cross Index of the Biologi- cal Abstracts to locate information. As class numbers' are used to designate subjects. the former can also be coordinated in place of the latter for information retrieval. The analysis of subjects into their categories or facets. and the subsequent operation of combining terms from different facets to form a compound sub- ject preceds the linear coordination of class numbers to designate it. With the UDC as the base of the present operation. it is possible to replace keywords of uniterm indexing sy s terrr by class numbers. Let such a system be named as 'Class index'.

Being a proposition on information retrie- val through UDC class numbers. it takes advantage of the absence of rigidity in their citation order which allows free interpolation of facets. and the ease with which facets can be introduced by special auxiliaries and taken apart by facet indicators.

The intra-faceted r el at ions h ip is so manifest in a composite class number that use of any other device for inter-linking them may be considered a superfluity. The facet indica- tors with colon symbols may take c a r c of the syntactical problems of uninflected English language. but constituent class numbers re- presenting individual facets at the post-coordi- nation stage will pose a problem in a complex

Vol 18 No Z June 1971

1.f Bose

Publications &: Information Directorate (C. S. I. R.). New Delhi-IZ

subject with mul tiple central relationship of concepts involving inter-related disciplines.

An unconventional scheme of using numerals as symbols may be experimented upon to bring 'roles' into use. We can, shelve the problem of 'weights' for the present as they. only signi- fying preferences. have not much bearing upon the problem of IR (informatiou. retrieval) as yet.

The idea of class coordination originates from underlying principles of uniterm and cross index. but as the former is too well-known to be elaborated upon. we may examine the latter index.

Cross Index

Each volume of Biological Abstracts.

published by the B'io s cierices Information Ser- vice. contains B. A. S. 1. C.• Author Index.

Biosystematic In dex and Cross Index at the end. They se rve as key to the list of abstracts and help to locate the desired information.

Cros s (Computer Rearrangement of Subject Specialities) Index is pre c ede d by an nAlphabetical Listing of Subject Headings"

containing approximately 500 subjects - all arranged in an alphabetical order. A total of 552 headings and subheadings appear in the Biological Abstracts (1971). These subject headings, being the same as the Section Headings. are repeated in the same sequence in the alphabetical listing of aubj e ct headingo of the Cross Index. Under these subje ct headings, reference numbers are cit~d in ten vertical columns according to the last digit of the number to facilitate cornp ar is o n of nurnb e re under more than one heading - thus enabl iIlg one to locate single reference relating to several subject categories. Obviously the pattern is the same as the uniterm index. But there is a small difference. The Cross Index is, in fact, a subject coordinator which helps to locate abstracts on any particular topic by matching reference numbe rs under two or more subject headings. The abstract numbers, com-

79

mon to these headings. will point out to the material containing information pertinent to the combination of the subjects matched. In uni- term index. keywords - supposed to desc ribe the contents of a document - are coordinated.

while in cross index. only headings. signifying the subjects dealt in them. are matched. We are aware that subjects can be denoted by class numbers. So. the coordination of class num- bers instead of subject headings or subjects can be tried to locate information. Let this re- trieval system be named as 'class index". Re- presenting subject headings by corresponding class numbers will entail all the advantages of a 'standard language' of a classification

schedule. and also to get around some of the problems encountered in uniterm and cross index. Uniterm index can operate successfully if the keywords are chosen not from the title alone. but also from the text and beyond so as to describe the subject contents of a document adequatel y.

Bioresearch Index of Biosciences Infor- mation Service also contains Biosystematic Index. Cross Index and BASIC besides a biblio- graphy. Additional keywords are provided to augment th e search. Obviously the idea is to present makeshift indicative abstracts in tele- graphic language with a profusion of supple- mented keywords so that bibliographical cita- tions would not fall short of an abstracting service. Example:

Bioresearch Index. Vol. 6(No. 5) May 1970.

3001 Machta L: Wind pollution and the wilderness.

Living Wilderness. 33(10) 1969.

Keywords: Winds. Pollution. Wildernessl Plants. Trees. Toxicity. Sulphur dioxide. Smog.

Now. coming back to the Biological Abstracts.

we see that since the subject headings are generally le s s apc c ific than concepts e x- p res s ed by keywords. their coordination may not p rovide any definite clue to specific docu- ment. So. the crOBS index is us e d in conjunc- tion with the BASIC to locate specific informa- tion.

Since subje ct headings with their nume- rous s ubhe ad ing s present a rather diffused scene. it b e c o rn e s necessary to guess potential keywords b e fo r e consulting the BASIC. The abstract numbers. thus located in BASIC. are then coordinatt!d with th o s e under the relevant subject head ing s of Cross Index. Obviously.

this inf'o r mat io-i -r.~trie·;","l.~ pr o c e s s pr c s uppo ae a

that subject hea d ing s confo rrn ing to a parti- cular topic s ho uld be visu alised before actuti coordination.

The fact that the Cross Index has also to be used in c o n iunctic n wi.th BAStC exposes it e inadequacy as an independent tool of informa- tion r etr te val., ·.~he BASIC its e lf ;.e not a e elf - suffic ient tool of

.m.

Moreover. the results obtained by matching cross index with BASIC have to be further coordinated with the Bioeystematic Index (or IR by searching according to a particular taxonomic category. The latter associates major taxonomic categories with the nature of the study. and gives the concurrent reference number(s). But the Bioeyetematic index plays only a secondary role because a taxonomic affil iation is made for every organism reported in the lite rature published in Biological

Abstracts and Bioresearch Index. whether the organism is tl;,e primary subject of study or not.

The fact that the Cr o s s Index is not a comprehensive tool for lR cannot be denied.

Its inadequacy is s ought to be remedied by KWIC index of BASIC. But this KWIC index.

based on keywords of unedited titles. alao faUs to score in many cases.

The

proposition

Suppose someone wants information on

"tr e atrn ent of marijuana addictions'.

A search ccn du c te d with the help of different type of inde xe s of BiDbgical At-s- tracts to locate the information yielded the following re s ultsi>

Cro~1l In dex;

Psychiatry - Addiction (;ndudcs alcohol.

d.:-,-,gs. s m oking etc.) 31 abstracts numbers are listed under this heading. No specific ~n~ormation could be

obtained. Moreover, one has to surmise that the subheading Drug addiction belongs to the main heading Psychiatry.

BASIC:

The KWIC index of BASIC contains only 2 keywords i,e , , Drug, dependence. The vital keyword, Treatment has not been utilized.

Biosystematic Index:

Primates Hominidae Psych Addict. By coordinating these 3 types of indexes, we come across abstract No. 102009 which gives the requisite information.

These indexes are not always successful in locating an information. In this particular case. the word treatment. and names of dif- ferent types of drugs including marijuana are not 1isted in these indexes. They indicate the abstract No.1 02009 aided by guesswork. One can identify the abstract with the needed infor- mation by reading the contents of the abstract.

The particular abstract. occurring in the Bio- logical Abstracts. Vol. 51(N0.8) September 15,

1970, is reproduced below:-

102009. Freedman. Daniel (Dept.

Psychiat. Univ •• Chicago.

nuusx).

(Patterns of human drug dependence, hallucinogens, ma r ijuan a , amphetamines ana narcotic addiction as a problem for society is discus sed).

In class index. we may analyse the same subject into multiple keywords which will constitute input to the system at pre-coordi- nation stage:-

UDC Nos Abstract Nos Drug addic^tion

Hallucinogens addiction Amphetamines addiction Narcotic addiction Treatment. Therapy

178.8 178.8 178.8 178.8 615

102009 102009 102009 102009 102009 Cards containing class index entries will have the same pattern as uniterm index with 10 vertical divisions (0-9) containing accession numbers of the listed documents. This will be

Vol l8 No 2 June 1971

known as the store or subject profile. These index cards will have the following entries:-

Class index entries Abstract Nos 178.8

178.8 178.8 178.8 615

Amphetamines addiction Drug addition

Hallucinogens addiction Marijuana addiction Treatment. Therapy

102009 102009 102009 102009 102009 Let this problem be now considered in the perspective of class index. At the initial stage of search technique. the que ry will be analysed into 2 keywords i,e .• Marijuana addiction, Treatment.

Following the principle of citation order of facets from the least to the most restrictive, these keywords will be aligned as follows:-

Marijuana addiction: treatment.

The composite class number of the subject.

consisting of 2 facets will be 178.8:615 Index cards with the class numbers 178.8 and 615 will now be scanned. The search will yield the following index entries:-

178.8 Marijuana addiction.

615 Treatment.

Both these entries are likely to contain several abstract numbers including, of course, the number 10200'~. This abstract number will stand out after the index entries are coordina- ted to pinpoint the document.

The scanning of index entries presents no difficulty because all entries bearing the same class number are filed alphabetically according to terms cited again8t respective class numbers - the latter remaining the same for any particular topic.

Let another example be cited. An infor- mation is needed on 'behaviour of rhes us monkeys under alcohol'.

Before consulting cross index. one has to surmise that the subject ill pharmacology with neuropharmacology and psychopharmacology all i',s cubh eadirig s , But coordination of entries under them yield no specific information, BASIC. however, provides some clue. The initial keywords 'Rhesus monkeys' and 'alcohol' of KWIC have a common abstract no. 125328.

!:II

The oic<Jystcmdtic Index also provides the following data:

Primate ';-{ominidae Ph a rm Psycho 125328

'I'he r-el evan t ab et rac t, con tain ed in th e

oi(:.',og:-::d Ab str ac t. Vol. 5 J(No. 22) Nov ern be r 1:-J 177':;·: 1..~'as follo'vVs:-

~2532e, Ch arn ov e, A.S. a.'1dH.F.

H2_!"~ .r»: }?:t'lDate Res Cent.., rJr..iv. Wi~_,

t..·i.;::.d';.)'.":'!), i.Vis~, USA. Exaggeration 0f oelf-

ii.gr? -3$ t,O:), fol.l o wing ale ohol i:.:.ge8 tion in rhe sus n;'),,~c".!ys, .7. A;)NO~M PSYCHQ!., 75(2): 2,0"'- 2C'9. i::..l',;,,:. 1770.

::.Ee.~f-directed aggressive behaviour vr».:

.;·~t.:or.dcd. in 11 rhesus mac ac ue s , .t:"'('l.tr ::Jub-

.~'f:'(.,~.:~wer e given 5-1511.1 ethyi alc ob.ol in var ic.us

-:.~o:>::,r:~'~~~"t.1;1~:·'.ti':.'.'n3;the others se rv ed a.s ccrit r ols . ,'3~\('':;l.,:~,~.r~ssicnincreased with ~:1).".rv.:r~,a3e in t~\.-::c.!:·l~~.U"'\.tof alcohol ingested ~ino.cpend.ent of

(~"),·i.:::!' (~:C' rcduc e d flu id ir..ta.ke. t., cor-t r as t,

~:,.lr:·;"'~··l::-.':.~.::.•..:;;be en reported to a~~.•?;l;J.::·r~.d..? self- .:":...·J.:.:.~..3.t.\:.1g b~ha..viours 0£ sr..~.:.dT-;:n~.

::"o0king back, we riotice that BASIC con-

:·',c::'.; :;, key^{wo r-ds} i,^{e , exagg e r at} i^{on ,} self-

,."';~'(~'~" 2Jco'101., ingestion, rhesus,

,!\?~ar~nt1y all the keywords, excepting

".1•.:r/:"'.~.l. rhesus. monkeys, are not helpful in

·c'-:;::-,h:-.g the i!.).formation. But this is one of the

·3.:',:y,+,:',cl;:~of the KWIC. The redundant kev-

"..,.,.-.:!c; in BASIC which tend to cre at e noise can

':,:,:-::lim,i;'1,:'\tec~y its c oo r diriat ion with. cross

in an I~ system bas e d on cl.as s c oo rdina-

;.~,out to the system for •.

,,-i,,"

.~.~·:l.:.("?,-;/'c~.ology

.

~(',,;"(~: "-".r\'....•'\eys

6~6~:3 613.~~~. 5°9. '".:

1.2538 1;~538 12531}

'.,.1.-:t; ~~'.:~: •. ~!""-

Alcohol is rn Rhesus monkey- Neuropsychology

12538 ),2538 12538

'I~e ~c:!a.:-cbtec~mi<p ...1.~ irvvolvee sur rrri s izig eo! ."'~,l;.~;:!.bJ..ei<:ey"-v·):':"~..:"'~o o":"!.{".':'?')'\'-::' the subj e ct, The

user profile. in terms of pre-coordinated class numbers. will be as follows:

Neuropsychiatry: rhesus monkey: alcoholism 616.8: 613.81: 599.9

Individual entries bearing these class numbers will be taken out of the file. and coordinated to specify the abstract (12538). As this indexing system bypasses coordination of terms with all their word variants. it can be said to have a direct approach to the problem of IR.

Like all other systems. a discipline has to be incu lc a.t e d while using this method of IR.

The auxiliary notation of distinctive symbols of UDC enables compound or complex subjects to be designated. classified and retrieved in terms of their component features which are often independently classifiable. Thus the need for repeated enume ration of similar con- cepts in different subject fields is substantially reduced. As the use of special auxiliaries without the benefit of A/Z indexing tends to hide the information, coloning and multiple entry should invariably be used in this system.

This is necessary because class numbers. and not keywords. form the access to subject pro- file at the search s tage, Thus facet indicators

notational de vi.ce s for introducing subse- quent concepts - will replace special auxilia- ries (both - or .0). An example may be cited here. The fol l o wirrg abstract appears in Nuclear Science Abstracts.

Vol. 24(No. 21) Nov J 5. 1970:-

46006. Determination of temperature in liquid metal-cooled nuclear reactor. Bulee v , N. 1.; Levchenko. V. E.; Polosukhina. K. N.;

Sholokhov A. A. At Energ. (USSR): 29:6-11 (July J970) (in Russian).

The citation order of the facets and class nuxnber will be as under:

Liquid metal-cooled nuclear reactor:

temperature measurement.

621.039.524.2.0346.6. 564.3

But in class index. this composite class num- ber will be cited in the form of its following constituent components for subsequent class indexing and post-coordination at the search stage:-621.039.524.2.034.6:621.039.564.3.

Again. consider the following document.

1498. Roll control with wing-tip jets.

629.7.062.22.025.84.036.53.

In cl as s index. the facets will be taken apart. and the document will have the following index entries:-

629.7.062.2 629.7.017.22 629.7.025.84 629.7.036.53

Altitude control systems Roll. lateral stability Wing tip

Compressed air or gas jet 1498 1498 1498 1498 Here. the facet indicator 629.7 has been re- peatedly used to replace the special auxiliary point nought (. o). Similar will be the case with hypen (-) auxiliaries also. Acce s s is the same for each element (or facet) at post-coordina- tion stage. But at the pre-coordination stage.

access to other elements (distributed relatives) other than the first cited one is less direct. So the use of facet indicators to break apart a composite class number incorporating special auxiliaries into its constituent elements is necessary for providing a ccess to them at the post-coordination stage.

To infuse flexibility for meeting varied requirements from UDC. a strict facet formula has not been imposed upon it. But in order to maintain its logical approach to the problem.

it is imperative that the citation order of facets in an integrated UDC number should preferably follow the principle of increasing concreteness at the pre-coordination stage to facilitate formation of a logical outline of the' subject. This consistency of approach in citing facets in a composite class number will be helpful both at the pre- and post-coordination stages. We may cite a few examples here to clarify the point.

Nuclear Science Abstracts. Vol. 24(No. 19}

October 15. 1970. contains the following abstract: -

4099&. (DP-Report- 265) Plutonium fuel cycles in high temperature gas cooled reactors. Project Dragon. Gratton. C. P.

(Atomic Energy Establishment. Winfrith.

England} Oct 1960. S8p. Dep. CFSTI.

(The pe r£ormance of various fuel cycle s using plutonium in high-temperature gas- cooled reactors are compared. Fuel costs for the various fuel cycles are analysed. )

Vol 18 No 2 June 1971

The terms which can represent the sub- ject along with their respective class numbers are as follows:

Gas-cooled reactor: Plutonium fuel: Fuel cycles: Performance

621.039.52.034.3:621.039.543.6:621.039.

516.4: 62.004.15

Input to the system will consist of the following index entries:-

621.039.52.034.3 621.039.543.6 621.039.516.4 62.004. 15

Gas-cooled reactor Plutonium fuel Fuel cycles Performance

40996 40996 40996 40996 Thereafter. the search and retrieval at the post-coordination stage and subsequent retrieval operation will follow the usual p roce- dure as described before.

Now. to illustrate the use of roles. con- sider the following subject:

Propulsion reactor for rockets and inter- planetary craft 621. 039. 578:629.19.

Class index entries using roles will be as follows:-

621.

oas,

Prpulsion reactor Rocket

Interplanetary craft.

Here the numerals (4 and 8). separated from the class numbers at their ends by semi- colon. represer.t roles. The use of this semi- colon is an arbitrary method intended only to distinguish roles from the run of numerals in a class number. The purpose is to achieve a homogenous look for class numbers with roles.

These roles (4and 8)have been taken from the Engineering Joint Council System.

A number of secondary periodicals pub- lish UDC classified abstracts. They include Irrigation &: Power Abstracts (India). J'ap an Telecommunications Review (Japan). Tropical Abstracts (Netherlands). World Fisheries (Ital y). and many othe1'5 from diffe rent count ry.

Languag e s differ but their contents are ade- quately exp r es s ed by UDC numbers. Some of these class numbers are co-extensive enough to denote five or more facets of a subject. This they define a subject as effectively as multiple keywords or unit concepts.

83

~~!e may now conclude the proposition by

citing one example from the Japan Telecom-

""lunicati.ons Review. Vol. 12(4) Oct 1970.

'JDG 621. 395. 5:621. 315. 23:621. 315. 17:62l.

315.2:677.73

Jel.ly-Fi.Ued Cable and New Undergrou:o.d D;~trF.:)U,i.c'n System, 18<,.0Nag aku ra..

., . 'Ial.~? No, 4(1970) pp. 273-278.

('~b':)'-,ghu.nderground d istribut ion fc.cili-

+.:(~2,ar « ux-ged frorn an aesthetic point of view,

~h,~C·-',":".-'~ntsystem using p ap e rcin sul at e d..

h.ad-'3heathed cable and han dhol e s , has such

".·.c<""O::vantagesas a high failure rate. unbe aut i-.

':·.•.1. ~.~'r=inations and high plant costs.

ThB new underground systems with the ie;:',cu!"'?!' of di.rect buried cable joints and jelly- fLUe.d -::i!.ble and of te z-rn inal vless distribution boxes, has been developed and is now under cornrne rcial, test.

This artist describes the features of j-::11y-fUled cable and outlirres the new system).

Cla ss Index entries will obviously be as :fo_l.1.o"'-·R;~

621.395.45 SZ1.315.23

Telephone transmission systems Underground cables. arrange- ments. laying.

Design. erection and maintenance of overhead lines.

Cables (insulated. sheathed and armoured conductors)

Conductor cable 621. 315.17

(,21. 315. 2 677.73

In this system. such secondary periodi- c al s with classified entries would have made :~.e task of. indexing. and their subsequent re-

·r.,.:'v;:;,} ",aeier .. but for the fact that one would '~.''>~;:J.'N"-YS agree with. cla ss numbe r s assigned.

tc .~';'.ny of these abstracts.

The trrocedure

T~~ procedu re is exce edinglv simple.

}'.~.~3·..1.?~~!c.tprof.i.te (cla.ssificd rUe) should be

~_\r!:·~·~.~ocly7.nt(.;-rpcs:ed with guide cards with ::--0.;.~'.'J.tc.1-.)s~.~dl.ca~ing a auccession ('f cl as si- 5ed subject zon e s , They will help one pick up I'ou.f;~t entries irrxmediatel v , Cui.d.e cards of.a

~r~:':'"t~.clJ,Ia.rsl_~.b5~ct;'~lJne v;jJ.l p resent the follow- ::.}S; l"01-".>

621. 43 621. 438 621. 45 621. 455 621. 47 etc.

Internal combustion engines.

Gas turbines.

Jet engine s , Rocket engines.

Solar energy engines

Here. class numbers have brought similar subjects together which would have been widely scattered by A-Z arrangement in an uniterm ind.ex file.

First of all. the topic on which informa- tion is sought will be mentally analysed into possible facets cited in least to most restric- tive order. and then outlined by a succession of class number elements - also aligned in that sequence. This empirical class number will be jotled down. Thereafter. the subject profile will be scanned. and index entries with concurrent class numbers be taken out of the file.

An operator working with this system will soon be familiar with subject zones and their respective class numbers of a file. This happens due to mnemonic value of most of the classification schemes e. g •• D.C•• U.D.C ••

C. C. etc. The rapid processing of uaeep r o- file and locating information is no longer a time consuming affair.

Noise will be minimised because index cards in the immediate vicinity of a sought one are likely to display various shades of the same subject enabling the operator to select ent r ie s more relevant than what he had in his mind or been seeking. Besides. when multiple central.- Iy related concepts of equal value are involved in a complex subject. links will be uaed with class number elements. The presence of link8 with class elements will alert the operator and prevent false coordination.

Besides simplicity of operation. the system has other advantages also.

Advantages

Being based on the UDC. the system will naturally inherit its advantages and limitation", Due to its general approach which treats all knowledge as an integral pattern of more or less related concepts. the UDC can designate all subjects - compound or complex. The flexibility of notations of UDC. which a Hows numerous alternatives in the d'''''''ification of

a subject field. is an asset to it. Although UDC is a precoordinate system. major alter- native approaches to the collection can be monitored by multiple entry (permutation). The great majority of sought coordinates in a given collection can be anticipated by permutations.

and thus most of the benefits of postcoordina- tion can be obtained. A subject profile based on classified sequence in undoubtedly prefera- ble to alphabetical alignment of word or sub- ject indexes wherein many like subjects.

scattered by A-Z arrangement. can only be retrieved through a maze of varied cross re- ferences.

In UDC. the absence of rigidity in not having a standard facet formula for citation order facilitates combination of class compo- nents in any desired manner. The roles will put the components into their proper perspec- tives irrespective of their order of alignment in the reconstituted class number.

Uniterm indexing essentially hinges on word indexing which has a lesser edge than subject indexing. Class indexing depends on coordination of class numbers denoting sub- jects. A class number. with implicity genus- species relationship not only identifies a sub- ject but also reveal its perspectives; and so indexing based on class numbers would seem preferable to word indexing. Class index.

possessing all the advantages that are heir to subject indexing. exhibit a vista of co- related information in its classified file. The main advantage of this scheme lies in elimination of homonyms. synonyms. near synonyms. over- lapping terms. vague terms. word ambiguities.

erroneous teems. superseded terms etc. by class numbers whose connotation does not variate with languages Or changed perspectives.

All these variants pose a problem to word indexing which. in turn. tries to control them through thesaurus. These term variants.

which plague word indexing. make it manda- tery to use multiple types of cross references.

They are presented in a systematic manner in the form of a thesaurus. But the vast expanse of existing subjects are also expanding to keep pace with the widening frontiers of science and technology. Int e r^er el at ed and overlapping dis- ciplines make the compilation of thesurii en- veloping all subjects and keeping them uptodate by additions and modifications a difficult p r o- ject.

Vol 18 No 2 June 1971

No such problem of prior compilation of thesurii will be encountered in the present system. Terms represented by given UDC class number in different national editions are claimed to be near as possible synonymous.

This is due to an extensive international co- operation in the preparation of schedules and in their maintenance.

Published and maintained uptodate by an international organisation (the Fill). the rela- tive index of UDC schedules will serve both as alphabetical index and thesurii at the search stage. The UDC derives the name 'univer.al' due to its comprehensiveness and co-exten.i- vity in designating any subject - however complex it may be. Limitations. if any. faced by the UDC in this respect will. however.

accrue to the present system also.

Indexing is primarily based on words and subjects. Uniterm indexing comes under the category of subject indexing. Word (or term) indexing comprising of KWIC. KWOC etc. is qualitatively inferias to subject indexing. In the preceding pages. the limitations of cross index. which is a compute rised version of subject index, has been reviewed. This subject index. being somewhat abstract. has to depend on word indexing whose drawbacks are too obvious to deserve mention.

Besides. a diffusion of subject head. and subheadings of the cross index i8 a poor substi- tute for the finesse of the UDC schedule num- bers. One can. however. say that the bewil- dering array of 552 subject headings of the cross index for matching may puzzle a layman.

but a subject specialist will be at home in it.

The argument is untenable because simplidty should be the key of any IR scheme so that a student or a beginer with some rudimentary idea about the subject. and specialists with an advanced knowledge may locate needed infor- mation with an equal ease. The ability to combine apec if'ic and broad c en cept s in alter- nate searches is the main advantage of the

classified arrangement. The UDC schedule displays spectra of like subjects at particular places of its schedul e. Once a reade r has been guided to them I::>yrelative index. he can cho s e appropriate cl a ss e e from its vicinity (or logI- cal alignment and subsequent search.

Considering all these factors. one arrives at the inevitable conclusion that the proposed class index can surmount some of

8)

the problems encountered in word and subject Indextng , In short. advantag e s of an analytico- synthetic classification scheme. classified catalogue and concept coordination will accrue to this indexing system whose roots lie deep inside them.

Mechanisation

In the present race for computerised information retrieval. the system that does not lend itself to mechanisation is apt to be looked down upon as an obsolete one. The class index can be mechanised because its counterpart (UDC numbers). consisting of numerals. deci- mal digits. alphabetical device. punctuation marks etc. are amenable to coding and com- puterised manipulation.

Conclusion

The exposition. as presented here. is by no means exhaustive. An attempt has only been made to lay the foundation for an experi- ment into an unexplored domain. Much groundwork is necessary which. when syste- matically done. will open up a new horizon in the region of information retrieval.

REFERENCES

[1] Automatic Indexing: A state-of-the-Art report. National Bureau of Standards Monograph 91. March 30. J965.

[2] BOSE H: Special auxiliaries: a critical appraisal. Indian Librarian 1969.

24(2). 59-67.

[3] COSTELLO J S: Coordinate indexing.

Volume VII. Ed. by Susan Artandi.

Rutgers. State University. 1966.

[4] FOMIN A A: On the possibilities offered by UDC and other descriptor retrieval system - a comparative study. Ann Lib Sci Doc 1967. 14(4).

177-81.

[5] Guide to the Universal Decimal Classi- fication (UDC). B. S. 1000C: 1963 (FID No. 345). London. B81. 1963.

[6] MILLS. JACK: Universal Decimal Classification. Vol. I. Rutgers.

State University. 1964.

[7] SHARP J R: Some fundamentals of infor- mation retrieval. London. Andre Deutsch. 1965.

-

go

z

N

<-t

~

It -D-.I

CD -.I

621. 039. 52. 034. 3:621. 039.543.6:621. 039. 516.4 40996. (DP - Report -265) Plutonium fuel cycles

in high temperature gas c ooled reactors.

62.004.15 PERFORMANCE

FUEL CYCLES ^{6 7 8 9}

621 .039.516.4

1498

6 7 8 9

621 .039.543.6 PLUTONIUM FUEL

621.039.52.034.3 GAS-COOLED R£ACTOR ^{6 7 8} 9 1498

0 1 Z 3 4 5 6 7 8 9 1498

1498

629.7 .062.2 ALTITUDE CONTROL SYSTEM

o I

I

^I

^I

4099~

629.7.017.22 ROLL, LATERAL STABILITY

o I

I

I

I

I

I

I

I

I

629.7 .025.84

o

629.7.036.53 COMPRESSEDAIR OR GAS JET

o

,

WING TIP

7 8 9

6Z9.7.062.22.025.84.036.53

1498.' Roll control with wing-tip Irts.

()

l'

»

Ul Ul

Z o

M><