Darrell R. Raymond

From data representation to data model: meta-semantic issues in the evolution of SGML

SGML provides standard representations for documents, but as documents become more fluid, we will need standard semantics for them as well. The ability to manage change is a fundamental capability of any system that supports document semantics. We look at three areas important in change management: equivalence, redundancy, and operators. We show how these areas are implicitly addressed in SGML and SGML-based standards, and argue that more explicit consideration would be useful both for evaluating current standards, and for developing new systems for document semantics.

Hypertext by link-resolving components

In conventional hypertext, links are explicit connections between specific regions of a text. We describe an architecture that treats links as the outcome of responses to user pokes. In this architecture, a hypertext is a collection of link-resolving components, each interpreting a user’s request according to its own resolution algorithm. Each link-resolving component is a set of cooperative processes, communicating with a central network manager. When a user points at some location within a window, each link-resolving component is given a key derived according to a previously-stored algorithm; the link-resolving components concurrently update their displays according to their algorithms for resolving the keys. Multiple applications can easily share a common source and be invoked simultaneously, providing a concurrent browsing mechanism. Two example hypertext employing this architecture are described.

Hypertext and the new Oxford English Dictionary

An alternative to manual composition of hypertext databases is conversion from existing texts. Such conversion often requires careful analysis of the text document in order to determine how best to represent its structure. We illustrate some of the issues of conversion with an analysis of the Oxford English Dictionary.

Flexible text display with Lector

Lector, an X11 application for flexible text interaction, is described. Lector was originally devised as a simple, rapid formatter for the display of online text databases. It was then discovered that Lector exhibited useful capabilities for a wide range of text applications. The ways in which Lector distinguishes between content and tags in descriptively marked-up text and achieves flexible interaction are discussed. Its applications as a text previewer, database browser, code prettyprinter, and menu utility are also discussed.<<ETX>>

Characterizing visual languages

A better understanding of the visual character of languages is important in developing ones ability to exploit the human visual system. The author briefly outlines Goodmans (1976) distinction between notational and analog languages, and describes its use in developing the notion of syntactic and semantic density as the defining characteristic of visual languages. Several languages are evaluated for their use of density. He concludes that practical languages are most visually effective when their layout is constrained by an important semantic domain.<<ETX>>

Measuring the effectiveness of personal database structures

Abstract The increasing proliferation of electronic billboards, hypertexts, and other informal electronic databases necessitates effective tools for personal data structuring. An experiment was conducted to investigate subjective processes involved during structuring an online database. Ten subjects organized two hundred proverbs into hierarchical structures over four sessions and used their structures to solve queries. Structuring and retrieval activity in the online environment was markedly poorer than in a previous manual experiment. In both experiments retrieval performance was correlated to the level of distinction employed in the construction of categories.

Playing detective with full text searching software

Searching large text databases often resembles detective work. We explored this notion with an experiment in which subjects used powerful full text searching software to solve problems about the Arthur Conan Doyle story The Hound of the Baskervilles. The experiment was conducted in two parts: in the first part subjects attempted to teach themselves about the software using only the documentation; in the second part, subjects used the software to answer questions such as What brand of cigarette does Watson smoke? The experiment provided a great deal of feedback about the usability of the software and the documentation. Among the results that have wider implications are the need for better display of context, and a need for careful documentation of the characteristics of full text searching.

An Experimental Study of a Gestalt Based Database for Mug Shots

The paper describes a series of experiments testing a recognition based computerized mug shot database system (gestalt based) for police use. The prototype is compared to the more traditional feature based database system. The gestalt based database was shown to perform significantly better than the feature based database when the quality of the signals was improved. In addition, the gestalt based system outperformed the feature based system in terms of efficiency of each question and ease of use. Issues concerning the coding of the database and the use of a hybrid system are also discussed.