Resource Library: Multimedia

home *** CD-ROM | disk | FTP | other *** search

/ Resource Library: Multimedia / Resource Library: Multimedia.iso / hypertxt / msdos / hypdiss / chapt5 < prev next >

Wrap

Text File | 1992-04-12 | 19KB | 319 lines

CHAPTER V SUMMARY, CONCLUSIONS, AND RECOMMENDATIONS Summary This was an exploratory case study dealing with the design of access to information stored in a hypertext system. It investigated the transferability of traditional, proven, information access techniques and information control methods to the new hypertext information medium. No other studies had yet directly addressed this area. The research questions in the study focused upon the present and potential information access approaches of a particular hypertext information system implementation. The investigator selected the MaxThink hypertext authoring system as especially appropriate to this study. This system had the most sophisticated hypertext network building and information organizational modules of the various systems examined by the investigator. MaxThink system developer Neil Larson is committed to a practical information publishing orientation, since he is involved in production of a large, complex, and regularly published hypertext information system. The principals involved with this hypertext publication are committed to the approach of creating structured access to knowledge. They have not been sidetracked from this by undue focus upon hypertext software technology. The investigator worked from a constructed conceptual model of the variety of traditional information access approaches (See Appendix A) <app-a>. He also directly used the MaxThink authoring tools to develop a sample hypertext document of approximately one hundred files. This hypertext covered the topic of electronic mail and network message access in the Texas Woman's University Academic Computing Center. The sample hyperdocument was built to gain better insight into the information access and organizing features present in the latest version of the MaxThink authoring system. Examples or illustrations of screens illustrating the use or product of the access and organizational tools are attached as Appendix F <app-f>. ***> SCREENS NOT INCLUDED IN THIS HYPERTEXT VERSION <*** Conclusions Study findings were presented in detail in Chapter IV. This case study supported the preliminary hypothesis about the suitability of the hypertext software platform for providing traditional information system approaches. The study findings led to the following conclusions: 1. The subject hypertext system can directly emulate or provide interface to software providing nearly all traditional information access methods. 2. The subject system is especially strong in the areas of information organization, design of structured access, and ease of use. 3. The taxonomy network matrix is a powerful tool for the organization of information. The creation of multiple hierarchical paths to document content resulted in an effective and easy to use means of structuring information access. The MaxThink system's matrix outliner offered an efficient approach for building the complex hypertext networds. This is a unique hypertext authoring system concept, offered by no other hypertext implementations. 4. The subject system developers have chosen an "integrating" or gateway approach to achieve sophisticated methods of information retrieval. These designers have focused upon providing simple, effective, information retrieval tools, rather than trying to develop a wide variety of esoteric retrieval modules from scratch. Their basic, complementary, access systems include a hierarchical network taxonomy, a KWOC index, and limited string and text index searching. For more sophisticated information retrieval approaches, these men prefer the use of hypertext links to execute their own external programs or third-party software. They recommend the use of third-party software of good quality to provide specialized information retrieval functions such as thesaurus or controlled vocabulary maintenance, full text searching, and database- style retrieval. They have chosen to avoid the learning and development effort and expense required for producing these highly specialized packages. 5. The simple MaxThink implementations of string scanning and text index searching combined with and complemented the basic hypertext associative link mechanism. The search engine quickly produces link lists pointing to document location of desired terms. The simple search engine does not have sophisticated searching capabilities; it only does single term searching or Boolean AND searching of multiple terms. However, the documents in the dynamically created link lists additionally contain the embedded associative links. As well as providing direct text-searching access, the retrieved documents also contain links to guide the user back into the authored hypertext hierarchies. The serendipitous combination of specific term access and hypertext link guidance results in a powerful retrieval aid. Generalization to the Hypertext Medium The study implied several generalizations for application to the broader hypertext genre. The writer points out, however, that these points apply specifically to hypertext authoring systems used for information dissemination or information storage and retrieval. They do not necessarily apply to other hypertext system types, such as those used for cooperative project management or issue development, for intergroup communication, or for personal information management (Conklin 1987) <refs -conklin>. These general observations are: 1. Along with providing basic associative linking, hypertext authors should use complementary combinations of traditional information access methods and metaphors. The combination will achieve synergy in end-user retrieval effectiveness. 2. Effective information access control devices are important in the hypertext authoring process. These may include such methods as thesaurus control or controlled vocabulary maintenance, use of effective classification and hierarchical approaches, specialized indexing quality controls, hypertext link control (to avoid blind or erroneous link references), and labor efficiency devices. Achievement of these controls will provide consistency and predictability for the end users of the hypertext information retrieval system. The labor and production efficiencies will avoid the purely subjective, "brute force" approach so common in many hypertext authoring operations. 3. Hypertext system developers and author(s) should consider the effectiveness of interfaces for supplementary use of specialized third party software of good quality. No single system developer can hope to achieve consistent quality of modular development in the many highly specialized areas of information retrieval. 4. A logical, consistent, and well-planned hierarchical taxonomy is central to the organization of an effective hypertext information system. 5. A powerful network or matrix outliner is preferable to other commonly used link control methods, in dealing with the complexity of an evolving link network. Many writers with substantial experience in hypertext production or information system design recommend outliners to efficiently produce hierarchical taxonomy structures (Frisse 1988a <refs -frisse>; Horn 1989 <refs -horn>; Larson 1987b <refs 16 4>; Martin 1990 <refs 17 12>). Others report upon the substantial power of multi- dimensional or matrix outliners for application to network or information system building in both knowledge representation power, and labor efficiencies (Danielsen 1989 <refs -danielsen>; Larson 1987a <refs 15 24>; Dewire and Locke 1990 <refs -dewire>; Perez 1991 <refs -perez>; Urr 1991 <refs 23 8>). Many of the workstation or Unix platform versions of hypertext implementations use a "graphic browser" display, a visual depiction of linked icons to show the detailed network structure. The graphic browsers can usually be used by both hypertext system authors and by end users. However, the nature of the iconic approach by definition limits the amount of node content or contextual information which may be presented. For authoring purposes, the graphic browser is also limited in the manipulability and hierarchical inheritance properties of iconic nodes. The writer strongly recommends that hypertext developers consider the implementation of specialized outliners or matrix outliner modules as authoring aids within their systems. They have the choice of developing their own tools, or designing compatibility with existing high- quality products of this type, such as GrandView (TM), Houdini (TM), MaxThink (TM), PC-Outline (TM), or ThinkTank (TM). 6. An efficient method of transporting and translating existing conceptual structures or taxonomies is needed. Although much attention is paid to the possibilities for direct import of documents into hypertext format, there has not been equivalent attention to importing knowledge structures or representations. Björkland (1990b) has reported on the difficulties of trying to manually convert a hierarchical subject classification. Automated import and conversion of such classifications could form the basis for the quick construction of substantial hypertext information systems. Obvious examples could be systems based upon existing taxonomies such as the various Wilson Index subject headings, Medical Subject Headings (MeSH), Engineering Societies index thesauri, and other specialized existing knowledge domain systems. 7. Systems developers should also work on development of better hardcopy depictions of the complex linked network matrix. These would be valuable for use in system authoring, editing, evaluation, and maintenance. Recommendations for Further Research This study focused on a single hypertext system implementation, and used case study methodology. It is recommended that further research be considered in the following areas: 1. Comparative study of configurations and performance of the information access devices provided by various hypertext system applications. 2. Survey of authoring system components used in different hypertext applications, with evaluation of the cost-effectiveness or productivity achieved in different configurations. 3. Study of the effectiveness of various approaches for representation of the hypertext network matrix or lattice in both display and hard copy formats. 4. Study of automatic term extraction or topical category assignment to hypertext nodes, as a linking guide for hypertext authors. 5. Study of methods for effective import or translation of hierarchical or enumerative classification schemes into hypertext format representation. This study has identified the definite potential of using hypertext software technology in the construction of effective information retrieval systems. The study pointed out the advantages of using matrix outliner software as an editorial tool and control for building of complex hypertext networks. The prevailing opinion is that construction of sophisticated hypertext systems is constrained by labor-intensive, overly subjective procedures. The editorial efficiency of the network outliner approach presents a possible solution to this basic problem. It is hoped that future studies on the topic will further contribute to the information retrieval design concepts and editorial efficiencies used in creating effective hypertext information systems. Afterword Alschuler (1989) <refs 1 5> was quoted in Chapter II, expressing puzzlement and concern over the unsatisfactory quality of the separate demonstration hypertext projects which worked on converting the Proceedings of the Hypertext '87 conference to hypertext format. She noted poor organization of the systems, inconsistency of indexing and linking, and general confusion in the design or depictions in the final products. She was chagrined, since these projects were managed by leading hypertext developers. This writer suggests that the major difficulties reported are the result of lack of experience in designing, organizing, maintaining, and representing complex knowledge or document collections. In the writer's judgment, the lack of professional library or information science practitioner involvement in hypertext system development has caused much of the "lost in hyperspace" problem. Computer applications developers, no matter how competent they may be in their own professional skill areas, are simply not trained in the methodologies of knowledge representation, information access design, information access system control mechanisms, and concise expression of subject organization and content. Information navigation or access problems have historically been empirically and theoretically addressed, by traditional library and information retrieval professionals. These specialists have also worked in both manual and automated environments. Their proven information access and control methods are the results of hundreds of years of trial-and-error approaches. Effective control systems for library catalogs, indexes, and database production operations already exist, and are obvious examples of approaches to this area. Existing systems already efficiently organize and handle enormous knowledge bases, such as Medline, OCLC, Dialog, and Chemical Abstracts index systems, Most current hypertext knowledge bases are trivial in size, compared to these systems. Roland Hjerppe, of the Swedish LIBLAB hypertext project, makes a telling point about the synergy to be gained from applications of traditional information technology to information retrieval methods: In general, I do not see traditional information retrieval techniques/tools and hypertext as alternatives but rather as complements. In other words . . . I would not contrast a hypertext approach with a traditional, but rather compare what a hypertext facility could add to a traditional tool and conversely what a traditional facility could add to a hypertext tool. (Hjerppe, Internet message to the author, October 1990) This is a rational view of the advantages to be gained from hypertext. The simplicity and ease of use of hypertext associative linking can effectively deliver information retrieval leverage to end users. There is much to be gained by combining the power of this new tool with the existing toolkit inventory of traditional, proven information access methods.