DRAFT Automatic Creation of Lexical Knowledge Bases : New Developments in Computational

Text processing technologies require increasing amounts of information about words and phrases to cope with the massive amounts of textual material available today. Information retrieval search engines provide greater and greater coverage, but do not provide a capability for identifying the specific content that is sought. Greater reliance is placed on natural language processing (NLP) technologies, which, in turn, are placing an increasing reliance on semantic information in addition to syntactic information about lexical items. The structure and content of lexical entries has been increasing rapidly to meet these needs, but obtaining the necessary information for these lexical knowledge bases (LKBs) is a major problem. Computational lexicology, which began in somewhat halting attempts to extract lexical information from machine-readable dictionaries (MRDs) for use in NLP, is seeing the emergence of new techniques that offer considerable promise for populating and organizing LKBs. Many of these techniques involve computations within the LKBs themselves to create, propagate, and organize the lexical information. 1 Introduction Computational lexicology began in the late 1960s and 1970s with attempts to extract lexical information from machine-readable dictionaries (MRDs) for use in natural language processing (NLP), primarily in extracting hierarchies of verbs and nouns. During the 1980s, NLP began reaching beyond syntactic information with a greater reliance on semantic information, locating this information within the lexicon. After reaching a conclusion (in the early 1990s) that insufficient information could be obtained about lexical items from MRDs, new techniques have emerged to offer considerable promise for populating and organizing lexical knowledge bases (LKBs). An underlying reason for the realization of these techniques seems to be the increasing capability to deal with the large amount of data that must be digested to deal with the overall content and complexity of semantics. This discussion begins with the assumptions about large amounts of information in lexical entries and particular computations made with this information in NLP. From this starting point, the paper describes emerging techniques for populating and propagating information to lexical entries derived from existing information with the LKB. The primary motivations for extending lexical entries comes from a need to provide greater internal