SP500207 Site Report for the Text REtrieval Conference chapter P. Nelson National Institute of Standards and Technology Donna K. Harman even if this means delaying some advanced or exotic modules. Our philosophy is that simple programs, well executed, will always out-perform complex tools poorly done. System Architecture ConQuest uses pre-built indexes to perform text database searches at fast speeds. In such a system, all text to be searched must first be indexed. The indexes built in this process can then be used by the text search engine to produce results. Both indexing and search use a dictionary with a semantic network to perform various NLP tasks. Queries Resufts Figure 1 ConQuest System Architecture There are other modules in the ConQuest system not shown in Figure 1. These include the library manager, which is responsible for system parameters, database configuration, resource allocation, and physical partitioning of the indexes. Also the dictionary editor, which can be used to edit words, meanings, links, and definitions. The Dictionary ConQuest uses a dictionary augmented with a semantic network to perform indexing and queries. The dictionary is a list of words where each word contains multiple meanings. Each meaning contains syntactic information (part-of-speech, feature values), and a dictionary defmition. The semantic network contains nodes which correspond to meanings of words. These nodes are linked to other related nodes. Relationships between nodes are extracted from machine re[OCRerr]dable dictionaries. Some example relationship types include synonym, antonym, child-of, parent-of, related-to, part-of, substance-of, contrasting, and similar-to. ConQuest uses the dictionary for morphological analysis (see below) and idiom processing. The semantic networks are used to expand the query to include related terms. Since connections are made between meanings of words, both in the dictionary and the semantic networks, processing is much more accurate compared to a simple thesaurus. 289