The document retrieval query language described here is based in part on the work done by the Maestro project, which is concerned with the development of techniques for managing structured text. The paper addresses the problem of document retrieval based on both document structure and document content. The language is in two parts--definition and a query.

MacLeod proposes a large number of language constructs based on Standard Generalized Markup Language (SGML). The paper describes the object types to be worked with (archive, group, document, element, and attribute) and then presents the general syntax of the language for both document definition and document query. The query language allows the user to specify a retrieval query based on text, document attributes that may not be part of the text of the document, and document structure (title, sections, paragraphs, sentences, and so on.)

Some of the author’s examples are confusing. In at least one instance, it is not clear which parts of the example are syntax and which are user-supplied (it appears as though some user entry items were inadvertently boldfaced).

The paper is divided into seven parts: “Introduction,” “Overview,” “Object Types,” “Search Capabilities,” “Document Creation and Deletion,” “Implementation Considerations,” and a “Summary.” The fourth section is by far the largest. The author also includes a list of references and an appendix containing the complete language syntax (divided into definitions, statements, selection, and operations). The paper would have been easier to read if the author had not been so creative with his user-supplied names. While the paper is not overly lengthy, the author presents all of the proposed language constructs. As a result, the explanations of what each construct means or is intended for are sometimes unclear. The paper could also have benefitted from a more consistent and logical presentation of the constructs.

Given the increasing interest in document management, document retrieval, and imaging systems by the major vendors, languages such as that proposed by the author will be of increasing interest as users gain the capability to store whole documents and then attempt to retrieve them.