Classification, Indexing and New Technology

(Unpublished paper, 1987)

Timothy C. Craven
School of Library and Information Science
The University of Western Ontario
London, Ontario N6G 1H1
Canada

Abstract

The topic of new technology's implications for indexing and classification almost always seems to imply computers, either directly or through various associated technologies. Major areas for consideration are: automatic indexing and classification; artificial intelligence; logic languages and other newer kinds of computer languages; graphics; microcomputers; online searching, and electronic publishing.

Introduction

The implications of new technology for classification and indexing are an extremely broad topic, and this article will necessarily be selective and summary in its coverage. Inspiration is drawn from a one-day conference held in 1986 in conjunction with the 43rd Conference and Congress of FID (23). A number of aspects not raised at this one-day conference will, however, also be considered.

At the present time, new technology as it relates to indexing and classification almost always seems to imply computers, either directly or through various associated technologies. Thus, the topic is very largely that of how indexing and classification are affected by computerization. The effects may be greater in indexing, or verbal subject analysis, than in what is traditionally called classification (44); but there are effects in both areas.

Technological advances other than those clearly associated with computers should not be entirely ignored. One instance that might be cited is voice indexing, a technique used to provide intellectual access to audiotapes for the blind. In this technique, key words are audible as normal speech in the fast forward mode, while the full text is what is heard if the tape is run in normal playback mode (1).

Computerization

The greater power and efficiency provided by various degrees of computerization has lead to information handling systems that are both more diverse and more complex. Complexity and diversity may create problems for users, and the power of computerization may in turn be applied to assisting users with these problems. System diversity may complement the diversity of users: computer power may be applied to customizing for particular user needs.

In theory, computerization makes possible much more personalized activity by information workers and users while at the same time allowing for sharing of data. The standards for the shared data are likely to be quite complex, though initially there may be a number of partial, competing standards and, in any event, many workers and users may never be aware of the full complexity. We can expect that a large number of levels of competency will be recognized and provided for in serving users, in classifying jobs, and in educating information workers (21).

String indexing (22), like citation indexing (61), provides an example of how computer power has encouraged the development of particular types of indexing systems. It also shows how further increases in computer power open up new possibilities.

In a typical string indexing system, each item to be indexed is described by a string of terms, usually with additional coding symbols. From each string, computer software generates a set of index entries, each entry under a different access point. When sorted and merged, and possibly enriched with cross-references, such index entries may combine to form an index display of considerable elegance. PRECIS is a well known example of such a string indexing system.

It was computer power that originally made sophisticated string indexing practicable. It did this by taking the burden of generating, sorting, and displaying multiple index entries from single descriptions off the shoulders of human indexers or clerical staff. Continuing increases in computer power are now making possible the further step of customizing string index displays to meet the needs of individual users and uses (18) (21) (22).

It should not, of course, be thought that all relatively complex indexing schemes are a response to the capabilities of the computer. Take, as an example, Farradane's Relational Indexing, a scheme for representing structures of syntactic relationships between terms in document descriptions, the aim being to provide better retrieval of technical documents. Although eminently suited to computerization, this was originally a manual scheme and was in fact computerized only after many years (36).

Nor has computerization always resulted in an increase in sophistication. On the contrary, the initial impetus may be to simplify in order to make things easier to mechanize. The Library of Congress Subject headings, for instance, in their manual form, have used various punctuation marks to bring together similar types of subject heading. The new American Library Association filing rules, on the other hand, influenced by the desire for simple mechanization, pay no attention to punctuation (44).

Computer power has facilitated the production of various tools to assist the indexer. A simple example is the index-in-reverse, a tool that gives the index terms assigned to each indexed item (34). In the 1980s, indexers and abstractors for a database producer typically sit at terminals and type their output directly into the database. Checking of spelling and verification of indexing terms are often built into the system (64).

Production of controlled vocabularies such as thesauri has been facilitated by the availability of computers. Basic editing, validation, and entering of inverse semantic relationships can obviously be accomplished automatically. In addition, some RT relationships can be generated automatically from hierarchical relationships (63). In one string indexing system, POPSI, a hierarchically organized thesaurus can be created automatically from descriptions of indexed items (28). Computerization also allows controlled vocabularies to grow more readily in response to new concepts, because indexers are aware that any additions can easily be standardized at a later time (44).

The efficiency and accuracy of input by indexers using a controlled vocabulary may be increased by appropriate input technology. A light pen can be used to point to descriptors on a screen display (12). Alternatively, especially if the controlled vocabulary is brief, it may be printed in a form suitable for optical character recognition (43).

For the most part, the needs of the Third World will not play a substantial role in future developments. Indeed, the general adoption of new subject analysis systems and the deemphasizing of printed materials at the expense of machine-readable data may be an actual disadvantage to less developed countries. Nevertheless, computer technology may encourage the taking on of the task of providing efficient subject access for speakers of languages for which indexing systems are not already well established (44).

Most computers have been designed to deal with text in the Roman alphabet as well as with numbers. Languages using other alphabets, such as Arabic, Russian, and Korean, have thus been at some disadvantage historically. Still more difficulties have attended languages such as Japanese or Chinese that use syllabic or ideographic writing systems. Special equipment may be acquired or developed where volume warrants; but, in other instances, some scheme of romanization, whatever its inadequacies, may have to be adopted (68).

The ready access to personal information provided by computerized systems has led to legislation designed to protect the privacy and accuracy of such information. Some British indexers have been concerned about the extent to which one such piece of legislation, the Data Protection Act, will affect indexing operations (26) (70).

Automatic indexing and classification

One dream long inspired by the computer is, of course, that of completely automatic indexing or classification of documents, with no need for human intervention. While computerization led to early experimentation with automatic classification (8), the most notable early successes were with simple kinds of extraction indexing. The earliest title-based computerized extraction indexes date from the 1950's, the best known being Luhn's KWIC index (50). Automatic extraction indexing is the basis of much online free-text searching today.

The relative merits of free-text searching and searching using controlled indexing and classification schemes continue to be probed (73). There are both theoretical and practical problems to the use of the natural language of free text in searching. For example, in an evaluation of the use of the STAIRS system for full-text retrieval of documents from a database of about 350,000 pages for legal support, recall averaged less than 20% (5) (6). Vocabulary control is thus still favored by many.

Automatic assignment indexing involving vocabulary control is a more ambitious aim than simple extraction indexing. The indexing of the 1971-1977 ABI/INFORM database may be cited as an example of automatic assignment indexing in a practical and fairly general setting. Here, the computer matched terms in each article's title and abstract with terms in a bridge vocabulary and then entered appropriate authorized terms or field codes in the index term field (76). Simple automatic book indexing using a thesaurus has also been suggested: a program searches the text for each occurrence of a term or its associated strings and creates an entry to the index when either is found (30). Of course, if a supporting thesaurus or classification scheme must still be developed and updated by hand, the indexing or classification is not purely automatic.

Experiments with automatic classification have tended to be based on surrogates, such as abstracts or tables of contents, rather than on the complete texts of the original documents. This has been especially true when the original documents are of book length (35). In any case, sophisticated automatic classification of documents has tended not to proceed beyond the experimental stage. The main problem facing automatic systems has been cost, which may be solved by future increases in computer power. Given a solution to the cost problem, some contend that automatic systems would perform no worse than manually assigned vocabularies (69).

Artificial intelligence

Using computers to simulate human intelligence in various ways falls under the general heading of artificial intelligence. While artificial intelligence also includes robotics and machine perception, the areas with especial impact on indexing and classification seem to be knowledge representation, natural language processing, and expert systems.

In terms of knowledge representation, the predicate logic, semantic nets, and so on of artificial intelligence and related fields can, to some extent, be paralleled by traditional indexing and classification devices such as roles, facets, and thesaural relations (78). The main limitation that prevents information retrieval thesauri from being considered as knowledge bases in the knowledge representation sense is their dearth of procedural information; that is, of active assistance in applying the factual information that they contain (46).

Natural language processing includes both the analysis and the generation of texts. Information retrieval has been seen as relating mostly to analysis, whether of documents or of queries. Much less often has it been seen as relating to text generation. Nevertheless, research on automatic text generation (60) may be significant for indexing in the future. An automatically generated text might be automatically indexed or classified not from the text itself, but from the database from which the text is generated.

Most work on expert systems in relation to libraries relates to helping end users to search online databases. A pilot expert system has, however, been implemented to assist MEDLINE indexers to assign subject terms more consistently, and it has been suggested that the capabilities of this system might be extended to proposing changes to the MeSH vocabulary (46). Work has also been reported under way on an expert system to assist catalogers in cataloging maps according to the Anglo-American Cataloguing Rules (8).

Computer languages

Much artificial intelligence research has been facilitated by the development of special computer languages, especially so-called logic languages like Prolog and LogLISP. A logic language is partly a computer programming language and partly a formal language in which truth relations can be expressed. It is beginning to be suggested that such languages be applied to tasks in information retrieval (33). Predicate calculus, which underlies much of Prolog, has been proposed as an approach to defining certain indexing language concepts in a formal way (66).

If a logic language such as Prolog is to operate efficiently on a large database, a classification scheme of some sort is needed in order to select the parts of the database that will be processed. A faceted scheme, based on the ideas of Ranganathan, has been used experimentally in structuring a database for this purpose (23). Given the usefulness of classification, the relative value of adapting existing schemes and designing new schemes to suit new technology is a question for discussion.

Another important type of computer language is database management languages, which form parts of many database management systems. Database management systems are widely used for a variety of applications, including indexing. Some, such as INMAGIC and BASIS, are specially designed with features that make them suitable for retrieval from bibliographic or textual databases.

Three models of data are widely used in database management systems: hierarchical, network, and relational. Most document retrieval systems may be viewed as following a fourth model, the inverted model. But the relational model has some advantages when one wishes to take account of additional data such as thesauri (24) (52) (53) (54). Its use in document retrieval might involve the storage of indexing tools such as stoplists, stem dictionaries, and thesauri as relations and their manipulation using relational operators (25).

In addition to logic languages and database management languages, other newer types of computer languages may also have special usefulness in indexing and classification. Object-oriented languages, for example, view the inside of the the computer as consisting of a collection of "objects", each belonging to a particular class; the objects exchange messages, and each object performs certain functions when requested. An object-oriented programming language such as Smalltalk might be used as the basis for interactive thesaurus editing and accessing (47). Spreadsheet software has, somewhat similarly, been suggested as a possible model for online browsing of concept and terminological systems (62).

Graphics

Quasi-spatial representations are one way to orient people to the vast amount of information in a large computerized database, and graphics technology can be a major contributor to providing such orientation (4). Use of graphics may be seen as capitalizing on the spatial functions of the right brain in addition to the logical and verbal functions of the left brain (77).

As graphics relate to indexing and classification, they seem mainly applicable to showing structures of links between terms or concepts. Computer graphic display of concept links was first mentioned as a possibility in 1962 (32). Concept links of interest to indexers and classifiers may be divided into two main categories: semantic, or paradigmatic, on the one hand and syntactic, or syntagmatic, on the other. Semantic links are the kinds typically handled by thesauri and classification schemes; syntactic links are the kind typically expressed by roles, facet indicators, prepositions, and the like.

Interactive graphics software has been used to produce displays of term relationships for a thesaurus, and graphic displays of term relationships may be generated automatically from machine-readable thesauri (9) (12). A number of advantages have been claimed for such graphic displays: giving an overview of the environment of a concept, avoiding dispersion of related terms, improving indexing exhaustivity and consistency, improving retrieval, helping the nonspecialist who does not understand indexers' conventions, and aiding in thesaurus construction (4). Such displays need not be limited to precontrolled vocabularies, but may also be used for statistical word associations in full-text databases (4).

Automatic graphic displays of syntactic relationships may be used as an aid in understanding complex descriptions in string indexing (15). Using certain experimental software, indexers can edit source descriptions directly as graphically displayed networks of terms (16) (17).

Physical three-dimensional models of term relations have been suggested and actually constructed (67). Such models might theoretically be simulated on high-speed interactive graphics terminals, where users could have the illusion of being able to view a concept network from different physical angles. Whether such an arrangement would be of practical value has yet to be determined.

The subset of graphic systems that fall under the heading of videotex present particular problems because of the small amount of text that can fit on the screen at any one time. A tree of menus was the approach taken initially in designing videotex retrieval systems, because designers needed to obtain quickly something that worked. Later, the possibility of adding other access methods such as keyword searching (71) alphabetical subject directories, and cross-references (81) was considered. The problem of how to optimize the design of the tree-structured menus, given the screen-size constraint, has also been addressed (82). Early interest in videotex systems has now declined, however, and this does not seem to be an important area of current research.

Microcomputers

Microcomputers have obviously made computer power available to many who had little access to it before. For individual indexers, for example, a wide variety of software packages are now available to assist in the clerical aspects of indexing work (37).

Microcomputers allow a high degree of quick interaction between the human user and the machine. A microcomputer-based screen editor, for example, may permit an indexer to type in input that is free from obvious coding errors (19). It is even possible for an indexer to appear to work directly on one part of a final index, while, at the same time, changes are automatically being made to other parts of the index to ensure consistency (20).

Microcomputers often have strong, and relatively cheap, graphics capabilities. Tree-structured online graphic displays of thesaurus relationships are available in at least one microcomputer-based information retrieval system (41).

Documents of a quality that previously required conventional typesetting will more and more be produced using "desktop" equipment (3). This opens up production of high-quality printed indexes to a much larger group of potential index producers.

Online searching

The ways in which indexing and classification will be used clearly have implications for the kind of indexing and classification that should be done in the first place. The growth of online searching is thus a major way in which computers have influenced, and will continue to influence, indexing and classification. Whether the searching involves discipline-oriented databases or library catalogs, similar considerations apply. It has been found that users of a variety of online information retrieval systems, including online catalogs, encounter much the same problems and many of the same factors underlie their behavior (7).

Modern online public access catalogs allow for many more access points than did the traditional library card catalog. The enrichment of catalog records with extra descriptors, say from tables of contents or indexes, has been suggested and experimented with. An increase in the average number of regular subject headings per item has already been observed (29). The first generation of online library catalogs emphasized recall; catalog databases are continuing to grow, however, and, as they do, precision in searching becomes more and more important (51). Multiplication of access points can contribute to both.

The growth of online searching has caused abstracts to be seen more explicitly as a source of index terms. In a recent survey of abstracting guidelines, nearly half included instructions aimed at free-text searching. Most treated abstract terms as complementary to controlled descriptors; some, however, called for embedding descriptors in the abstracts. In either case, the abstractor was seen as fulfilling, at least in part, the role of an indexer (39).

It has been argued that, because indexing is fundamentally indeterminate beyond a certain point, the indexing language will inevitably cause problems, and the searcher must be assisted if retrieval is to be effective (2). Moreover, increased availability of computer power has meant that more and more end users are doing online searching (64). Ease of searching or assistance with searching thus become all the more important. To help searchers with the indexing and query languages, alternative approaches that may be adopted include search intermediaries, natural retrieval languages, and intelligent interfaces (27).

New capabilities of microcomputer equipment allow for three-party data communication. Such communication can simplify access to online databases by allowing the end user either to consult online with a human search expert while performing an online search or to communicate at a distance with an expert intermediary who is performing the actual search (75).

Computerized search interfaces have been available for some time (48) (55) (56) (80). Some, such as BRKTHRU, reside on the host computer, and some, such as IN-SEARCH, on microcomputers (31). Most use mapping algorithms that are based on text characteristics, such as word frequency or statistical associations (38). Some are identified as expert systems: for example, IT or USERLINK, DIALECT, EXPERT, and CANSEARCH (27). All have limitations in what assistance they will provide (27). Nevertheless, to the extent that they are not concerned with problems of query languages, all of them can compensate to some degree for weaknesses at the indexing stage.

A common weakness at the indexing stage is use of different vocabularies for different databases. One mechanism proposed to provide greater compatibility among indexing vocabularies is a vocabulary switching interface between the searcher and the database. A number of such interfaces have been experimented with, including Batelle's VSS. Switching vocabularies may also be used to assist in selecting the appropriate databases to search (10), or as an aid in free-text searching (65). If vendors find the benefits sufficiently substantial, they might in future be willing to charge nominally or not at all for use of such an aid (65).

New commands and operators have been added to online search software to make it more helpful to searchers. One example is the ZOOM command that allows frequency-ranking of index terms from a sample of a document set (59). The searcher is thus assisted in selecting additonal terms for refining a search strategy in a way that does not require a preexisting thesaurus. Proximity operators, originally designed for full-text searching, have tended to displace more powerful, if somewhat more difficult, syntactic devices such as links and roles in controlled vocabulary retrieval.

Online retrieval has been based mainly on verbal indexing, but a number of uses of classification schemes may be envisioned: improving recall and precision; saving time in keying in search terms; contextualizing vague search terms; providing a structure for browsing; providing a framework for representation and retrieval of nonbibliographic data; and serving as a switching language (72).

The first use of a classification scheme for online searching and browsing was demonstrated in the AUDACIOUS project in 1967-1968. The scheme used was the Universal Decimal Classification, which was already largely in machine-readable form (14). It is only more recently that the Dewey Decimal Classification has been investigated as an aid to online browsing: the 19th edition of this scheme was the first to be produced by computerized photocomposition and hence to be machine-readable. Among points to be noted from this investigation is the value of tagging parts of compound notations (57) (58) (79).

The present Library of Congress classification scheme has a number of weaknesses for online browsing: nonhierarchical notation; nonavailability in machine-readable form; size; lack of an up-to-date unified index; and difficulties with auxiliary tables. On the other hand, for online searching in general, it has the important strengths of specificity, relative uniformity of use from library to library, and application to a large number of items (11).

A technology such as CD-ROM offers the possibility of whole databases' being purchased for local searching, as they have been on tapes since the mid-1960s. It has been suggested that such use of CD-ROM could kill little used databases, because the loss of revenue to online search services from major databases would force up prices (80). If so, certain materials might cease to be indexed, or might be indexed only by larger, less specialized services.

Electronic publishing

Increased computer power allows for the storage of more and more text online. The resulting quick availability of full text does not necessarily eliminate dependency on indexing (64). Yet the kind of indexing and classification required when the original is easily retrieved may be different from what is required when it is not.

Bibliographic control of online publications is difficult for a number of reasons, including the ease of changes or deletions in the publications, the vagueness of the distinction between published and unpublished material, and the lack of an effective legal deposit system (83).

In 1983, the Association of American Publishers initiated a project aimed at producing a uniform standard for identification and tagging of elements such as title, subtitle, and main headings in an electronic manuscript (83). Such standardization of full-text records will aid in including them in databases as well as in indexing and classifying them, both manually and automatically.

The structure of electronic publications should take advantage of the special features of online retrieval systems. The hierarchical approach to document structure has advantages for online searching when compared to the alternative narrative and autonomous approaches (42) (49). New attention may have to paid to indexing systems, such as chain indexing, that are specifically oriented toward hierachically structured information stores.

In addition to text, the storage of more graphic and other visual information will become economical (40), and methods of indexing, classifying, and retrieving this type of information may become more important.

Conclusion

In summary, new technologies affecting indexing and classification almost always mean computers. Computer power has encouraged the development of new indexing systems, such as string indexing and citation indexing, and continues to open up new possibilities, especially in the area of customizing output to meet specific needs.

New tools have been made available to help indexers. At the same time, simple automatic indexing of free text is widely used in information retrieval systems. More sophisticated automatic classification or indexing methods tend to remain experimental.

Artificial intelligence research with implications for indexing and classification generally falls into the areas of knowledge representation, natural language understanding and generation, and expert systems. Newer types of computer languages, such as logic languages and database management languages, can be applied to indexing and classification tasks. Improved graphics technology may help indexers to visualize structures of conceptual relationships. Microcomputers have put far more computer power into the hands of individuals, both indexers and searchers.

Computerized searching has changed the indexing environment by making more access points available for the indexed documents. Searchers can also receive various kinds of online assistance. Classification schemes will need to be adapted or developed for use in online browsing and retrieval as well as in more traditional applications like arranging books on shelves.

Finally, electronic publishing means text and graphics will be more readily available, and organized in somewhat different ways from print. These changes in document availability and organization should be taken into account in designing future indexing and classification systems.

References

1. "Book talks recipes". Feliciter. 27 (2): 3. 1981.

2. MJ Bates. "Subject access in online catalogs: a design model". Journal of the American Society for Information Science. 37 (6): 357-376. 1986.

3. D Bawden. "Computer output devices: a tutorial review". Journal of information science. 11 (1): 1-8. 1985.

4. S Bertrand-Gastaldy; CH Davidson. "Improved design of graphic displays in thesauri - through technology and ergonomics". Journal of documentation. 42 (4): 225-251. 1986.

5. DC Blair. "Full text retrieval: evaluation and implications". International classification. 13 (1): 18-23. 1986.

6. DC Blair; ME Maron. "An evaluation of retrieval effectiveness for a full-text document-retrieval system". Communications of the ACM. 28 (3): 289-299. 1985.

7. CL Borgman. "Why are online catalogs hard to use? Lessons learned from information-retrieval studies". Journal of the American Society for Information Science. 37 (6): 387-400. 1986

8. H Borko. "Expert systems applied to library cataloging; FID/CR panel remarks, Sept 13, 1986". International classification. 13 (3): 154. 1986.

9. DF Cahn. "Computer-aided visualization of database structural relationships". American Society for Information Science proceedings. 17: 358-360. 1980.

10. AY Chamis. "The usefulness of switching vocabularies for online databases". American Society for Information Science proceedings. 22: 311-314. 1985.

11. LM Chan. "Library of Congress Classification as an online retrieval tool". Information Technology and Libraries. 5 (3): 181-192. 1986.

12. J Chaumier; P Fourteau. "Le traitement des thesaurus a schemas fleches par l'informatique graphique interactive". Documentaliste. 16 (1): 9-14. 1979.

13. PA Cochrane. Redesign of catalogs and indexes for improved online subject access: selected papers of Pauline A Cochrane. 1985. Oryx Press.

14. PA Cochrane; K Markey. "Preparing for the use of classification in online cataloging systems and in online catalogs". Information technology and libraries. 4 (2): 91-111. 1985.

15. TC Craven. "Microcomputer-generated graphic displays as an aid in string indexing". Journal of the American Society for Information Science. 31 (2): 123-124. 1980.

16. TC Craven. "A microcomputer-based notepad for graphic editing of concept subnetworks". American Society for Information Science proceedings. 18: 243-244. 1981.

17. TC Craven. "Editing of concept subnetworks with a microcomputer-based display system". Journal of documentation. 38 (1): 29-37. 1982.

18. TC Craven. Microcomputer customizing of permuted index displays. American Society for Information Science proceedings. 19: 63-65. 1982.

19. TC Craven. "A NEPHIS screen editor as an aid in permuted index generation". Journal of information science. 5 : 187-191. 1983.

20. TC Craven. "An experimental electronic worksheet for articulated index production". American Society for Information Science Annual Meeting proceedings. 22: 315-318. 1985.

21. TC Craven. "Changing technologies: impact on information: the case of string indexing". Indexer. 14 (4): 255-256. 1985.

22. TC Craven. String indexing. 1986. Academic Press.

23. TC Craven. "Classification, indexing, and new technology; report on 3rd regional FID/CR conference". International classification. 13 (3): 153-154. 1986.

24. RG Crawford. "The relational model in information retrieval". Journal of the American Society for Information Science. 32 (1): 51-64. 1981.

25. RG Crawford; IA MacLeod. "Modular indexing in a relationally based document retrieval system". Canadian journal of information science. 6: 67-75. 1981.

26. JE Davies. "Man bytes [sic] index and (maybe) index bites man--some notes on the Data Protection Act". Indexer. 14 (4): 249-253. 1985.

27. G Deschatelets. "The three languages theory in information retrieval". International classification. 13 (3): 126-132. 1986.

28. FJ Devadason. "Online construction of alphabetic classaurus: a vocabulary control and indexing tool". Information processing and management. 21 (1): 11-26. 1985.

29. SM Dhawan; AN Yerkey. "Trends in subject heading assignment in cataloging records during 1974 -1978". Information processing and management. 19 (4): 213-222. 1983.

30. M Dillon. "Thesaurus-based automatic book indexing". Information processing and management. 18 (4): 167-178. 1982.

31. TE Doszkocs. "Natural language processing in information retrieval". Journal of the American Society for Information Science. 37 (4): 191-196. 1986.

32. LB Doyle. "Indexing and abstracting by association". American documentation. 13 (4): 378-390. 1962.

33. CM Eastman. "The use of logic programming in information retrieval experimentation". American Society for Information Science proceedings. 20: 58-59. 1983.

34. C Edward. "New aids to indexing". Bulletin of the American Society for Information Science. 5 (1): 30-31. 1978.

35. PGB Enser. "Automatic classification of book material represented by back-of-the-book index". Journal of documentation. 41 93): 135-155. 1985.

36. J Farradane; D Thompson. "The testing of relational indexing procedures by diagnostic computer programs". Journal of information science. 2: 285-297. 1980.

37. LK Fetters. "A guide to seven indexing programs..plus a review of the 'Professional Bibliographic System'". Database. 8 (4): 31-38. 1985.

38. R Fidel. "Towards expert systems for the selection of search keys". Journal of the American Society for Information Science. 37 (1): 37-44. 1986.

39. R Fidel. "Writing abstracts for free-text searching". Journal of documentation. 42 (1): 1-63. 1986.

40. C Fox. "Future generation information systems". Journal of the American Society for Information Science. 37 (4): 215-219. 1986.

41. HP Frei; JF Jauslin. "Graphical presentation of information and services: a user-oriented interface". Information technology. 2 (1): 23-42. 1983.

42. TR Girill. "Narration, hierarchy, and autonomy: the problem of online text structure". American Society for Information Science proceedings. 22: 354-357. 1985.

43. P Halpin. [Personal communication]. 1983.

44. RP Holley. "The consequences of new technologies in classification and subject cataloguing in Third World countries: the technological gap". International cataloguing. 14 (2): 19-22. 1985.

45. RP Holley. "Classification in the USA". International Classification. 13 (2): 73-78. 1986.

46. SM Humphrey; NE Miller. "Knowledge-based indexing of the medical literature: the Indexing Aid Project". Journal of the American Society for Information Science. 38 (3): 184-196. 1987.

47. P Kleinbart. "Prolegomenon to 'intelligent' thesaurus software". Journal of information science. 11(2): 45-53. 1985.

48. LR Levy. "Gateway software: is it for you?". Online. 8 (6): 67-79. 1984.

49. MB Line. "Redesigning journal articles for on-line viewing". Hills PJ, Trends in information transfer. Greenwood. Pp.31-46. 1982.

50. HP Luhn. "Keyword-in-context index for technical literature (KWIC) index". American Documentation. 11 (4): 288-295. 1960.

51. CA Lynch. "Cataloging practices and the online catalog". American Society for Information Science proceedings. 22: 111-115. 1985.

52. IA MacLeod. "Towards an information retrieval language based on a relational view of data". Information processing and management. 13 (3): 167-175. 1977.

53. IA MacLeod. "SEQUEL as a language for document retrieval". Journal of the American Society for Information Science. 30 (5): 243-249. 1979.

54. IA MacLeod; RG Crawford. "Document retrieval as a database application". Information technology. 2 (1): 43-60. 1983.

55. RS Marcus. "An experimental comparison of the effectiveness of computers and humans as search intermediaries". Journal of the American Society for Information Science. 34 (6): 381-404. 1983.

56. RS Marcus. "Development and testing of expert systems for retrieval assistance". American Society for Information Science proceedings. 22: 289-292. 1985.

57. K Markey. "Class number searching in an experimental online catalog". International classification. 13 (3): 142-150. 1986.

58. K Markey. "Overview of Dewey Decimal Classification (DDC) Online Project; FID/CR panel remarks, Sept 13, 1986". International classification. 13 (3): 154-156. 1986.

59. WA Martin. "Methods for evaluating the number of relevant documents in a collection". Journal of information science. 6 (5): 173-177. 1983.

60. KR McKeown. "Discourse strategies for generating natural-language text". Artificial intelligence. 27 (1): 1-41. 1985.

61. E Miller; E Truesdell. "Citation indexing; history and applications". Drexel library quarterly. 8 : 159-172. 1972.

62. W Nedobity. "Terminology and artificial intelligence". International classification. 12 (1): 17-19. 1985.

63. A Neelameghan; R Maitra. Non-hierarchical associative relationships among concepts: identification and typology + semi-automatic method of preparing thesaurus for a specific subject field. 1978. FID/CR.

64. ML Neufeld; M Cornog. "Database history: from dinosaurs to compact disks". Journal of the American Society for Information Science. 37 (4): 183-190. 1986.

65. R Niehoff; G Mack. "The Vocabulary Switching System: description of evaluation studies". International classification. 12 (1): 2-6. 1985.

66. S Reball. "On the application of predicate calculus in information indexing and the degree of indexing languages formalisation". International forum on information and documentation. 3 (3): 14-17. 1978.

67. PA Richmond; NJ Williamson. "Three-dimensional physical models in classification". International Study Conference on Classification Research 3, Ordering systems for global information networks: proceedings. Pp.188-203. 1979. FID/CR; Sarada Ranganathan Endowment.

68. GH Rogers. "From cards to online: the Asian connection". Information technology and libraries. 5 (4): 280-284. 1986.

69. G Salton. "Another look at text-retrieval systems". Communications of the ACM. 29 (7): 648-656. 1986.

70. A Sandison. "Data protection and the indexer". Indexer. 15 (1): 24-25. 1986.

71. AH Schabas. "Videotex information systems: complements to the tree structures". International Study Conference on Classification Research 4, Universal classification I. Pp.285-291. Indeks Verlag. 1982.

72. E Svenonius. "Use of classification in online retrieval". Library resources and technical services. 27 (1): 76-80. 1983.

73. E Svenonius. "Unanswered questions in the design of controlled vocabularies". Journal of the American Society for Information Science. 37 (5): 331-340. 1986.

74. DA Thompson. "Interface design for an interactive retrieval system; a literature survey and research description". Journal of the American Society for Information Science. 22 (6): 361-373. 1971.

75. R Trautman; DL Graham. "Three-party telecommunications to facilitate public use of interactive systems". American Society for Information Science proceedings. 22: 258-261. 1985.

76. L Trubkin. "Auto-indexing of the 1971-1977 ABI/INFORM database". Database. 2 (2): 56-61. 1979.

77. RH Veith. "Information retrieval and spacial orientation". American Society for Information Science proceedings. 22: 250-254. 1985.

78. BC Vickery. "Knowledge representation: a brief review". Journal of documentation. 42 (3): 145-159. 1986.

79. AS Wajenberg. "MARC coding of DDC for subject retrieval". Information technology and libraries. 2 (3): 246-251. 1983.

80. ME Williams. "Transparent information systems through gateways, front ends, intermediaries and interfaces". Journal of the American Society for Information Science. 37 (4): 204-214. 1986.

81. NJ Williamson. "Viewdata systems: designing a database for effective user access". Canadian journal of information science. 6: 1-14. 1981.

82. NJ Williamson. "Videotex information retrieval systems: the logical development and optimization of tree structures in a general online interactive system". International Study Conference on Classification Research 4, Universal classification I. Pp.277-284. Indeks Verlag. 1982.

83. R Yamamoto. "Another interface: electronic publishing and technical services". Canadian Library Journal. 43 (4): 235-244. 1986.
Notes, 2002
This was an invited paper for The Electronic Library which the editor subsequently decided not to publish.

Home

Last updated January 24, 2008, by Tim Craven