Moving towards a Topic-Based DDC

Although oft regarded as a class-based system, many potential uses of the Dewey Decimal Classification (DDC) system are better supported by a topic-based view. This paper explores automated strategies for identifying equivalence and hierarchical relationships between Relative Index headings, one of the steps required to fully implement a topic-based data model for the DDC. Data used in detecting relationships beyond the small number of relationships already explicitly coded include qualifiers in headings, variant name notes, hyphenated words, acronyms, the structure of Relative Index headings, adjectival modification, equivalence and hierarchical relationships between Library of Congress Subject Headings, and synonymy and hypernymy relationships in WordNet.