1 “Based on word and phrase searching, full-text research is distinguished from manual research by the total absence of a priori indexing and its concomitant–-the interposition of third-party judgment between the researcher and the legal authority he seeks.” Jerome S. Rubin and Robin L. Woodward, LEXIS: A New Tool for Tax Research, 1 J. Corp. Taxation 42, 47 (1974). Jerome S. Rubin is president of Mead Data Central, Inc., the company that developed the LEXIS system.Google Scholar

2 Case law, on the other hand, varies as widely as does the writing style of individual judges. Ideas are expressed in no set format and in no carefully defined order. Moreover, the same idea may often appear in several different verbal formulations. Thus, decisional law can be defined no more precisely than as a series of words, each of which may draw its meaning and importance from the words around it. The lack of any formal characteristics for written judicial opinions poses the greatest obstacle to the design and development of computerized case law research systems.Google Scholar

James P. Chandler, Computers and Case Law, 3 Rutgers J. Computers & L. 202, 204 (1974).Google Scholar

3 This article presents a detailed explanation of the LEXIS system of Mead Data Central, Inc.; it is also relevant to such similar retrieval systems as JURIS, QUIC/LAW, LITE, and DATUM-all of which are described individually in American Bar Association Standing Committee on Law and Technology, Automated Law Research (Chicago: American Bar Association, 1973)-and the WESTLAW system of West Publishing Co. The discussion here and the conclusions reached do not apply to statistical systems (see text at 181) or to any other system that does not permit the formulation of document search requests in a “Boolean” language (see note 15 infra). The QUIC/LAW and WESTLAW systems are hybrid systems that permit either Boolean searching or statistical searching at the option of the researcher. I describe the WEST-LAW system in a short article that will be published in an upcoming issue of the Jurimetrics Journal.Google Scholar

4 A more complete list of the materials stored within the LEXIS system appears in the text at 190. Additional information on the LEXIS system and a complete list of documents (in “LEXIS Library Materials”) may be obtained from Mead Data Central, Inc., Ill West First St., Dayton, Ohio 45401; tel. (513) 222–9561.Google Scholar

5 West Publishing Co. has experimented with a headnote retrieval system for several years and has recently announced its availability. See Stephen Singular, Computers and the Law: The Revolution Is Almost Here, 5 Juris Doctor 53, 54–55 (1975).Google Scholar

6 See Richard M. Hull, Legislative Computer Applications: The Illinois Story, 3 Rutgers J. Computers & L. 187 (1974), and Durward P. Jackson, The Utah Legal Information Project (TULIP)-Findings and Recommendations (available from the Utah Bar Foundation, 425 E. First Street, Salt Lake City, Utah–-January 1975).Google Scholar

7 Mead Data Central, Inc., plans to establish entry points to the LEXIS system at libraries and bar association headquarters for use by attorneys not wishing to subscribe to the LEXIS service on a full-time basis. William G. Harrington, What's Happening in Computer-Assisted Legal Research? 60 A.B.A.J. 924, 928 (1974).Google Scholar

8 Great care is necessary in selecting such a list of words, as the following indicates: Difficulties with searching by words are illustrated in the following paragraphs taken from a newsletter put out by LITE … (“LITE” Newsletter. Volume 4, No. 3, July 1971, p. 3). “LITE” users are cautioned to refrain from asking for specific words or phrases unless they are absolutely certain that these words or phrases will do the job …. In a recent court martial, the user wanted a search run on the words “black light” or “mineral light” using these words no cases were found. Unfortunately he was interested in “fluorescent powder or paste” as used under an “ultra violet light.” The use of these words would have turned up seven cases on point ….”Google Scholar

Government of Canada Department of Justice, Operation Compulex: Information Needs of the Practicing Lawyer 49–50 n. 7 (Ottawa: Information Canada, 1972); reprinted in 2 Rutgers J. Computers & L. 188, 215–16 n. 7 (1972) [hereinafter cited as Operation Compulex].Google Scholar

9 Technically speaking, the researcher attempts to maximize “recall” and “precision” when he revises his search request. “Recall” is the ratio of the number of “relevant” documents actually captured to the number of relevant documents in the collection. Hence, a recall value less than unity indicates that some relevant documents were missed. “Precision” is the ratio of relevant documents captured to the total documents captured. Thus, a precision value less than unity indicates that some irrelevant documents were captured. See Gerard Salton, ed., The SMART Retrieval System–Experiments in Automatic Document Processing 39 (Englewood Cliffs, N.J.: Prentice-Hall, Inc., 1971). Unfortunately, the above definitions are useful only if some objective standard of relevancy exists. Research carried out at the American Bar Foundation by William B. Eldridge indicates that attorneys frequently disagree as to what judicial decisions are relevant to a particular legal problem. See William B. Eldridge, An Appraisal of a Case Law Retrieval Project, in Proceedings of the Computers and the Law Conference–1968, 36 (David L. Johnston, ed.; Kingston, Ontario: Faculty of Law, Queen's University, 1968). Because the phrase “relevant document” connotes an objective, consistent, and logical standard of relevancy, and since relevancy in this context appears to be at least partially subjective, I have avoided using the phrase in this article.Google Scholar

10 Under the current [manual] system, whenever a lawyer conducts in-depth research, he begins at a fundamental level by reading a text on the matter, then he may turn to a digest or encyclopedia to familarize himself with the history and evolution of the legal principle in question. Only after he has an understanding at this conceptual level does heturn to retrieving cases …. What will happen with a computer? If lawyers rely on it to the point that it replaces the basic reading and understanding necessary for the current method of legal research, then lawyers' ability to understand, to interpret, to integrate the law will be eroded.Google Scholar

Operation Compulex, supra note 8, at 22; 2 Rutgers J. Computers & L. at 216–17.Google Scholar

11 To familiarize oneself with a computer retrieval system is as yet no trivial matter. Although the search language may be mastered in an hour, as Mead Data Central and QUIC/LAW have stressed, considerable experience is necessary before one could skillfully handle any research problem that may arise in a lawyer's practice. Some participants estimated that at least one hundred real queries would be necessary to reach that stage …. This long training period can be explained in part by the virtual absence of systematic knowledge on search strategies, but should also be attributed in part to the process of developing “computer search habits and reflexes.”Google Scholar

Ejan Mackaay, Reflections on the First National Conference on Automated Law Research, 3 Rutgers J. Computers & L. 310, 322 (1974).Google Scholar

12 For more information on statistical systems, see Salton, supra note 9. See also, William B. Eldridge and Sally F. Dennis, The Computer as a Tool for Legal Research, 28 L. & Contemp. Prob. 78 (1963) and Eldridge, supra note 9. Eldridge and Dennis experimented with the use of statistical techniques for retrieving judicial decisions in a project that was sponsored jointly by the American Bar Foundation and IBM.CrossRefGoogle Scholar

13 See Gerard Salton, A New Comparison Between Conventional Indexing (MEDLARS) and Automatic Text Processing (SMART), 23 J. Am. Soc'y Information Scientists 75 (March 1972). By using statistical retrieval techniques in conjunction with a semi-automatically generated thesaurus and by designing the retrieval system so that it automatically alters its searching strategy in accordance with the searcher's satisfaction or dissatisfaction with the retrieved documents, Sal ton and his associates appear to have developed a statistical retrieval system (SMART) that performs better than a conventional human-indexed retrieval system (MEDLARS). The MEDLARS system is widely used to assist in the retrieval of documents relating to medical subjects.CrossRefGoogle Scholar

14 In this article, LEXIS operator words are capitalized, and words of text are lower case. When one types out a search request on the keyboard of a LEXIS terminal, one may type the entire request in upper or lower case.Google Scholar

15 Retrieval languages of the LEXIS variety are referred to as “Boolean” languages because the LEXIS operator words “AND,”“OR,” and “AND NOT” are essentially identical in their functioning to the intersection, union, and complement operators of Boolean algebra. Boolean algebra is the algebra of classes–it enables one to describe, in a mathematically precise way, what objects are included in what classes. For example, if a first class, A, includes all documents containing the word “estoppel” and a second class, B, includes all documents containing the word “contract,” then the rules of Boolean algebra would define the union of A and B as a new class including all documents containing either word, and the intersection of A and B as a new class including only documents containing both words. In the LEXIS language, the union of A and B is expressed as “estoppel OR contract,” and the intersection of A and B is “estoppel AND contract.” See appendix B of Irving M. Copi, Symbolic Logic (4th ed. New York: Macmillan Co., 1973). Retrieval languages of the LEXIS variety are not purely Boolean, how ever, because they typically include a number of non-Boolean operators. For example, the LEXIS language includes the non-Boolean, positional logic operators “W/-” and “PRE/-,” which are described above, and also the non-Boolean, relational logic operators “IS,”“AFT,” and “BEF,” which may be used to restrict retrieval to documents dated on, before, or after a certain date or between any two dates. See the discussion in the text and also LEXIS–A Primer 12 (Dayton, Ohio: Mead Data Central, Inc., 1975). The “BUT NOT” and “W/SEG” operator words are also not strictly Boolean, since they are effective only when the words which they conjoin appear within a single segment of a document.Google Scholar

16 “The possibility of bypassing important cases is a real one, since the user must supply keywords that exactly match those used by the opinion-writer.” Chandler, supra note 2, at 212. “Notwithstanding all these searching aids and no matter how thorough the search, the user has no guarantee that he has recovered all relevant authorities or even those that are most relevant. The user has 100% reliance that the computer will do what is asked of it, i.e., search and retrieve cases or statute law containing his search words. However, he cannot rely on the computer to retrieve all information relevant to the topic in which he is interested.” Operation Compulex, supra note 8, at 22; 2 Rutgers J. Computers & L. at 215.Google Scholar

17 “A serious problem with full text systems is the number of irrelevant documents that often accompany the relevant. A related problem is the tendency to output a formidable number of documents.” Operation Compulex, supra note 8, at 21; 2 Rutgers J. Computers & L. at 214.Google Scholar

18 “Proponents of computer systems maintain that legal research time will be reduced. This has yet to be demonstrated …. [T] here are … situations where the searcher can spend a great deal of time (perhaps even more than it would take to search manually) at the terminal before he is satisfied with the results.”Id. at 22–23; 217.Google Scholar

19 “[T]he researcher, without the aid of a manually-prepared index, determines which, and in what order and location, words or combinations of words are likely to occur in relevant statutes or court decisions ….” Legal Research and the Computer 4 (Dayton, Ohio: Mead Data Central, Inc., undated). See also note 1, supra.Google Scholar

20 These internal indices are word indices or concordances of the document text. The use of such an index is not a theoretical necessity, for the computer can read through each document every time a search is carried out, but to read through each of a large collection of documents becomes excessively costly when frequent searches are carried out. There are retrieval systems otherwise similar to the LEXIS system that do not use a word index. See, e.g., Robert Chalice, Donald Dillaman, and Lorraine Borman, RIQS Remote Information Query System (Jan. 1970), copies of which may be obtained from the Vogelback Computing Center, Northwestern University, Evanston, Illinois 60201.Google Scholar

21 Ill. Rev. Stat. ch. 43, sec. 135 (1973).CrossRefGoogle Scholar

22 31 Ill. 2d 393, 202 N.E. 2d 9 (1964). The other two cases are Jones v. DeWig, 25 Ill. App. 3d 423, 323 N.E. 2d 475 (1974), and People v. Stamps, 8 Ill. App. 3d 896, 291 N.E. 2d 274 (1972).Google Scholar

23 An edge-notched card system similar to the one used by the Patent Office is described in American Bar Association, Standing Committee on Law and Technology, Computers and the Law–An Introductory Handbook 2–4 (2d ed. Chicago: American Bar Association, 1969).Google Scholar

24 USCA, “General Index–S to T” (1971). Copyright 1971, West Publishing Company.Google Scholar

25 The USCA index fills eight volumes. It typically cross-indexes to a depth of two to three levels.Google Scholar

26 At present, a LEXIS terminal may be leased for about $500 per month, of which $220 is for the special terminal and $280 is for communication charges. Additional terminals cost less than the first. The LEXIS service itself normally costs full-time subscribers about $75 to $110 per hour of use, on the average. The precise charge per second is less at times when no search is in progress and more when the system is actually searching. Full-time subscribers may also take advantage of a lower “off peak” rate, about $60 per hour, by doing their research at certain times of day. Hence, a single-terminal station could be leased for a minimum of $500, per month, not including the cost of searching. Public terminal users are currently charged a somewhat higher hourly rate than full-time subscribers.Google Scholar

27 The approximate hourly charge for purchasing time-sharing computer services from Service Bureau, Inc., is about $12 per hour, not including the cost of leasing a terminal or any extra charge for document storage in the Service Bureau computer system. This would be the charge that an engineer would pay if he were using such a service to do mathematical computations, for example. A suitable terminal may be leased for less than $150 a month.Google Scholar

28 Mead Data Central's LEXIS system presently stores about 4.5 billion characters on disk storage units. Jerome S. Rubin, Update on LEXIS: A Progress Report, 15 Jurimetrics J. 326 (1975). Using standard IBM disk storage units of the type available in 1975, it would cost over $1 million per year just to store this much data, or roughly $5,000 per year per subscriber with 200 subscribers. Using IBM Model 2344 disk storage units, which IBM does not plan to introduce until mid-1976, it should be possible to store this much data for under $200,000 per year, or roughly $1,000 per subscriber, again assuming 200 subscribers.Google Scholar

29 The cost of retyping documents can vary from $0.50 to $2.50 per 1,000 characters, depending upon the quantity and quality and also depending upon where the typing is done. Retyping 535 million characters for the Ohio library at $0.75 per 1,000 characters would cost about $400,000.Google Scholar

30 The LEXIS system has close to 200 subscribers now, but the number of subscribers is still a small fraction of the potential market. See Rubin, supra note 28.Google Scholar

31 Optical memory systems, particularly those of the holographic variety, promise to lower significantly the cost of archival data storage in the future. See O. N. Tufte and D. Chen, Optical Techniques for Data Storage, 10 IEEE Spectrum 26 (Feb. 1973); O. N. Tufte and D. Chen, Optical Memories: Controlling the Beam, 10 IEEE Spectrum 48 (Mar. 1973); Jan A. Rajchman, Promise of Optical Memories, 41 J. Applied Physics 1376 (1970); L. K. Anderson, Application of Holographic Optical Techniques to Bulk Memories, MAG 7 IEEE Transactions on Magnetics 601 (1971); and Rein Turn, Computers in die 1980s 191–204 (New York: Columbia University Press, 1974).Google Scholar

32 This will result as a by-product of the trend toward the use of computerized document-composition systems. “The computer, through photocomposition techniques, has already made a contribution in the publication and revision of statute law.” Operation Compulex, supra note 8, at 18; 2 Rutgers J. Computers & L. at 210. See also references cited supra note 6.Google Scholar

33 Following three years of studying ways to apply computer technology to legal research, a group of Ohio lawyers in 1967 organized Ohio State Bar Association Automated Research Corporation (OBAR). To provide lawyers in Ohio with a full-text, computer-assisted legal research system, OBAR entered into an agreement with Data Corporation, an advanced technology company in Dayton, Ohio, under which its full-text information retrieval system (developed in 1964 for search and retrieval of Air Force reconnaissance engineering documents) would be adapted to the law. Data Corporation was subsequently acquired by the Mead Corporation, which incorporated Data Corporation's Information Services Division as a new wholly-owned subsidiary, Mead Data Central, Inc. MDC took over the contract with OBAR and began in 1969 to offer a limited number of Ohio lawyers a developmental, full-text, interactive legal information retrieval system.Google Scholar

Legal Research and the Computer, supra note 19, at 6.Google Scholar

34 The organizations are: OBAR (Ohio State Bar Association Automated Research Corporation), TEXLEX (TEXLEX, Inc., an affiliate of the State Bar of Texas), MOBAR (Missouri Bar Research, Inc., an affiliate of the Missouri Bar), and IBAR (Illinois Bar Automated Research, an affiliate of the Illinois State and Chicago Bar Associations). In New York State, Mead Data Central, Inc., entered into an agreement with the New York State Bar Association–no subsidiary organization was established.Google Scholar

35 NCAIR (National Center for Automated Information Retrieval) is a nonprofit, educational corporation chartered by the New York State Board of Regents. Its original name was Lawyers Center for Electronic Legal Research (LCELR), and it was formed in 1966 under “the general aegis and sponsorship” of the New York State Bar Association and its tax section. See Edwin M. Jones, The National Center for Automated Information Retrieval and Its Role in Electronic Legal Research, 15 Jurimetrics J. 79 (1974).Google Scholar

36 In general, this description of the LEXIS system is also applicable to any retrieval system utilizing remote-entry terminals (see supra note 3) and permitting the use of a Boolean-type retrieval language (see supra note 15). A comparative study of several such systems appears in Martin, The Features Legal Researchers Need in an Interactive Retrieval System (as Derived from an Analysis of Eleven Interactive Systems), in Ronald A. May, ed., Sense and System in Automated Law Research (Chicago: American Bar Association, 1975).Google Scholar

37 The full text of the documents is stored on the surface of spinning disks in the form of magnetic impulses. When asked to retrieve a specific document for display, the central computer system positions a playback head similar to those used in tape recorders over that portion of a spinning disk where the requested document is stored. The playback head recovers the magnetically encoded document text and transmits it by telephone to an attorney's terminal where it is written out on the terminal's built-in television display.Google Scholar

38 See note 4, supra.Google Scholar

39 For example, suppose one is searching through the New York LEXIS case collection for cases relating to collisions between bicycles and cars. The initial search request might beGoogle Scholar

automobile OR car AND accidentGoogle Scholar

but this request captures too many case decisions–4,352 to be exact. Rather than discard the results of this initial search, one might elect to add the following modification to the search request:Google Scholar

AND bicyclistGoogle Scholar

As modified, the request captures only four case decisions–too few to be very useful. A second modificationGoogle Scholar

OR bicycle OR tricycleGoogle Scholar

would expand the scope of the search once more. This time 60 case decisions are retrieved. It is easier and less expensive to modify a search in this fashion than to execute a new search identical to the modified version of the original search. See LEXIS–A Primer, supra note 15, at 13–14.Google Scholar

40 Computer scientists generally call such a word index or concordance an “inverted file,” since it may be regarded as an inside-out version of the document library itself. In theory, one could reconstruct a document library from its word index, since it specifies the position of every word of text. In practice, this usually cannot be done because certain common words are deleted from such word indices to reduce both the size of the index and the time required to do a search.Google Scholar

41 For a complete list of nonsignificant, nonindexed words see LEXIS–A Primer, supra note 15, at 19.Google Scholar

42 Requiring the word position numbers to differ by no more than three is equivalent to requiring the corresponding words to be separated by less than three significant words. For example, in the phraseGoogle Scholar

the words “doctrine” and “loquitur” are separated by two significant words, “res” and “ipsa,” but their word position numbers “31” and “34” differ by three.Google Scholar

43 Similar conventions also govern the evaluation of most computer programming language. See, e.g., the discussion on pp. 67–75 of Mario V. Farina, FORTRAN IV Self-Taught (Englewood Cliffs, N.J.$$ Prentice-Hall, 1966), which explains the conventions applicable to mathematical and logical expressions written out in FORTRAN IV, a widely used computer-programming language.Google Scholar

44 This example was based upon some originally suggested by Professor Layman E. Allen, of the University of Michigan Law School.Google Scholar

45 A full explanation of truncation and equivalents may be found in the operating manual for the LEXIS system terminal and also in LEXIS–A Primer, supra note 15.Google Scholar

46 A more detailed explanation of date searching appears in id. at 11–13.Google Scholar

47 Personal communication from the ABA staff member who is in charge of the ABA/LEXIS project and from various attorneys who subscribe to the LEXIS system.Google Scholar

48 Secondly, a legal concept or rule, unlike a factual occurrence, cannot adequately be described, and any retrieval system which imposes this requirement on the user is doomed to failure. At the very least, it is demonstrably true that no two lawyers, no matter how similar their training, background and outlook, will describe a given legal concept in the same way using the same language.Google Scholar

Philip Slayton, Electronic Legal Retrieval: The Impact of Computers on a Profession 22 (Ottawa: Information Canada, 1974), 14 Jurimetrics J. 29, 34–35 (1973).Google Scholar

If one is looking for an individual fluent in German, Russian and Italian with a degree in pharmaceutical chemistry, the [computerized personnel data bank of the Canadian government] … can readily respond with names of individuals meeting these qualifications. However, if one were looking for an individual fully at home with the German, Russian and Italian cultures yet who has no vested interest in any of these countries, it becomes an entirely different kind of question. Here one is no longer looking for data but for a personality, a history, in short, a concept; the computer has difficulty with such a question. Computerized legal information retrieval systems face the same difficulty for if they are to be of assistance to the lawyer they must be able to retrieve a concept in addition to data.Google Scholar

Operation Compulex, supra note 8, at 20; 2 Rutgers J. Computers & L. at 212.Google Scholar

Proponents of the abstract approach maintain that the only way a concept can be retrieved by computer is through the intervention of a human mind indexing the concept for the computer.Google Scholar

Id.; 2 Rutgers J. Computers & L. at 213.Google Scholar

If I have any regret, it is that we lawyers, courts, legislators, etc. are all working with what are frequently meager and inadequate symbols–words. As had been said, “words are the clothes which thoughts wear–only the clothes.” (Samuel Butler–Notebooks)Google Scholar

M. David Henkle, Automation, Law and Research, 14 Jurimetrics J. 270, 270–71 (1974).Google Scholar

49 See note 10, supra.Google Scholar

50 Gilbert v. Korvette, Inc., 457 Pa. 602, 327 A. 2d 94 (1974).Google Scholar

51 Id.Google Scholar

52 Only citations to decisions handed down since 1950 are considered here, since a computerized full-text collection of cases would be unlikely to go back in rime more than 25 years.Google Scholar

53 Pedretti v. Pittsburgh Rys., 417 Pa. 581, 209 A. 2d 289 (1965); Sevast v. Lancaster Yellow Cab & Baggage, Inc., 413 Pa. 250, 196 A. 2d 842 (1964); Ambrose v. Western Md. Ry., 368 Pa. 1, 81 A. 2d 895 (1951); and Evans v. Otis Elevator Co., 403 Pa. 13, 168 A. 2d 573 (1961).Google Scholar

54 Izzi v. Philadelphia Transportation Co., 412 Pa. 559, 195 A. 2d 784 (1963); Miller v. Delaware County Memorial Hospital, 428 Pa. 504, 239 A. 2d 340 (1968); and Banet v. Philadelphia, 226 Pa. Super. 452, 313 A. 2d 253 (1973).Google Scholar

55 Paul v. Hess Bros., Inc., 226 Pa. Super. 92, 312 A. 2d 65 (1973); Loch v. Confair, 372 Pa. 212, 93 A. 2d 451 (1953); and MacDougall v. Ford Motor Co., 214 Pa. Super. 384, 257 A. 2d 676 (1969).Google Scholar

56 403 Pa. 13, 168 A. 2d 573 (1961).Google Scholar

57 417 Pa. 581, 209 A. 2d 289 (1965).Google Scholar

58 These figures were obtained by searching for occurrences of the phrase “res ipsa loquitur” in the LEXIS system's New York, Ohio, and Missouri case collections.Google Scholar

59 A slightly different problem which might cause the bypass of key cases is the user's attempt to keep the volume of retrieved cases to manageable size by expanding the keyword to keywords or specifying that words appear in predetermined positional relationships. The addition of multiple search terms does not, and cannot, under present theory, preserve those cases which are relevant, but contain fewer than all the keywords requested by the user, or have them in a slightly different order than he has set out.Google Scholar

Chandler, supra note 2, at 212.Google Scholar

60 Texas Family Code sees. 14.02 to 14.04 (1974).Google Scholar

61 Texas Family Code sec. 21.03 (1974).Google Scholar

62 This search was carried out on the LEXIS system's Texas civil case collection in June of 1975.Google Scholar

63 517 S.W. 2d 669 (Tex. App. 1974).CrossRefGoogle Scholar

64 519 S.W. 2d 254 (Tex. App. 1975).Google Scholar

65 This search was also carried out on the LEXIS system's Texas civil case collection in June of 1975. The 40 most recent cases were studied to determine which were child custody cases.Google Scholar

66 An indexing system similar to that suggested here is presently used in the JURIS system. “JURIS is special in that it adds West Keys and other indexing terms to the documents. This can be done at the time of conversion but also after the document has been included in the bank and has been missed during an important search.” Mackaay, supra note 11, at 318.Google Scholar

The RECON-IV system of Informatics, Inc., Rockville, Md., also includes such a provision, according to Phillip Nyborg, director of Legal Information Systems.Google Scholar

67 Naturally, there are many different ways in which indexing can be accomplished. The LEXIS system engineers suggest that all indexing terms be placed within a special segment of each document, as is done with die catchlines in the LEXIS New York case collection. Then the ability of the LEXIS system to search within specified segments could be utilized to limit any search to the indexing terms or to the document text. Of course, a search not directed to a specific document segment would then be applied to both the indexing terms and the text. While abbreviations or number signs (see text at 213) would not be mandatory in such a system, they would still be helpful to keep the searcher from confusing indexing terms with textual terms in searches that were not directed to a specific document segment. Any suitable prefix may be used instead of abbreviations or number signs, for example, the letters “IX” may be used as a prefix to all indexing terms. Numbers may be used as indexing terms, but then a guidebook explaining what the numbers mean would have to be provided. Meaningful words, phrases, and abbreviations are far easier to remember and use than numbers. Computer programmers, for example, go to great lengths to substitute meaningful letter codes for meaningless numeric codes whenever possible in their programming languages.Google Scholar

68 Sec p. 211, supra.Google Scholar

69 At least the LEXIS system treats number signs in this fashion. In other systems, slashes, asterisks, dollar signs, or other such symbols might have to be substituted for the number signs used here.Google Scholar

70 There is one important advantage that a computer possesses over a library for intelligence amplification. This is in the area of fact retrieval. Few library systems are indexed for facts except in a limited way. A computer has the advantage that it is capable of storing the full texts of the cases, so that cases dealing with facts represented by descriptive words familiar to the user can be retrieved readily with a computer system but only slowly with a library indexing system. Consequently, in the fact retrieval area, the amplification factor of a man-computer team is much higher than average.Google Scholar

Reed C. Lawlor, Amplification Factor of an Information Retrieval System, 15 Jurimetrics J. 289, 301 (1975).Google Scholar

71 But one must be careful when one selects factually oriented textual words for use in formulating a document search request. See note 8, supra.Google Scholar

72 See p. 197, supra.Google Scholar

73 See pp. 211–12 supra.Google Scholar

74 I am not aware of any such retrieval guides prepared for the document libraries of the LEXIS system. Mead Data Central has prepared excellent technical guides to assist in learning how to use the LEXIS system, but technical guides do not make one familiar with the vocabulary of the set of documents or suggest specific formulations of document search requests developed by others to solve particular retrieval problems.Google Scholar

75 This particular breakdown of the material contained in article 9 of the Uniform Commercial Code is attributable to Professor Roy L. Steinheimer, Jr., Dean of Washington and Lee University School of Law.Google Scholar

76 In the JURIS system, the searcher may first have the main headings in an outline of some subject displayed. He may then have the subheadings under any desired main heading displayed. Sub-subheadings under any subheading may also be displayed and so on. Finally, the searcher may select an indexing term that appears opposite a sub-subheading. Using this indexing term, the searcher may have all documents retrieved to which the term has been assigned by an indexer. In the JURIS system, the outlines are all taken from portions of the West topical index, with West's permission, and the indexing terms are West key numbers which have been incorporated into the stored documents. B. W. Basheer, JURIS: Justice Retrieval and Inquiry System, in Automated Law Research, supra note 3, at 60–61.Google Scholar

77 The first project is an experiment in the use of the PLATO educational computer system in continuing legal education directed by Professor Peter Maggs, of the University of Illinois School of Law. Peter B. Maggs and Thomas D. Morgan, Computer-Based Legal Education at the University of Illinois: A Report of Two Years' Experience, 27 J. Legal Ed. 138 (1975). The second project is an experiment in using a minicomputer to conduct client interviews and to draft legal documents, directed by Henry Brown, of the Cook County Legal Assistance Foundation, Chicago. See Richard W. McCoy and William A. Chatterton, Computer-Assisted Legal Services, 1 L. & Computer Technology 2 (1968); Walter Rugaber, Computers May Cut Legal Costs, N.Y. Times, Feb. 17, 1974, at 83; and Directory of Jurimetrics Projects, 15 Jurimetrics J. 56, 57–59 (1974). I will soon publish a full description of the Cook County system. The third project, an experiment in the use of artificial intelligence to assist an attorney, is described in Walter G. Popp and Bernhard Schlink, JUDITH, A Computer Program to Advise Lawyers in Reasoning a Case, 15 Jurimetrics J. 303 (1975). Popp and Schlink actually suggest that their system, suitably modified, could serve as a front end for a LEXIS-type, full-text retrieval system.Google Scholar