Structural Methods for Lexical/Semantic Patterns

Structural Methods for Lexical/Semantic Patterns Scott A. Waterman Brandeis University Computer Science Dept. Waltham, MA 02254-9110 email: waterman@cs.brandeis.edu ABSTRACT This paper represents initial work on corpus methods for ac- quiring lexical/semantic pattern lexicons for text understand- ing. Recently, implementors of information extraction (IE) systems have moved away from using conventional syntac- tic parsing methods, instead adopting a variety of pattern based approaches for complex IE tasks. While there has been much work towards automated acquisition of lexicons for con- ventional syntactic processing, little progress has been made towards those for pattern systems, due primarily, in the au- thor's opinion, to a lack of a linguistic framework in which to view their use. In combining a functional view of both de- notational semantics and syntactic structure, this paper pro- vides a basis for examining the structural constraints between the two. This functional viewpoint is the starting point for methods to investigate the characteristics of the interaction between text and denotation, from the perspective of pattern- based systems. An approach for determining and exploiting these structural constraints is outlined in terms of building hierarchical lexical structures for text understanding. Exper- iment results for such a method are given, demonstrating the functionality of the approach. 1. Introduction. Recently, implementors of information extraction (IE) systems have moved away from using conventional syn- tactic parsing methods, instead adopting a variety of pattern based approaches for complex IE tasks (such as the MUC contests and the ARPA-sponsored TIPSTER research). These pattern-based systems use short and fairly specific lexical patterns to specify the relation be- tween strings in the source text and particular entries in a problem-dependent knowledge representation. This one-step process substitutes for a conventional two-level process of a full syntactic parse followed by semantic interpretation. With considerably less time and devel- opment effort (notably demonstrated by [11, 8]), these systems achieve performance comparable to more stan- dard systems that rely heavily on full syntactic analy- sis ([9, 5]). However, because these pattern-based sys- tems are still viewed as linguistically ungrounded and somewhat ad hoc, formal work in the application and acquisition of lexical patterns has lagged system devel- opment. In most current systems, patterns are produced through tedious hand analysis of text ([11, 4, 8]), while system coverage is gained either through extensive lin- guistic knowledge on the part of the researcher (in judg- ing appropriate pattern generalizations), or by generat- ing and testing massive numbers of patterns. One exception to hand analysis is Lehnert's work in [12], in which machine learning techniques are used to infer possible patterns for extraction. While this AutoSlog technique has dramatically reduced system development time, the inference techniques use only sparse linguis- tic information, provide no means of generalizing pat- terns across domains, and still require that the rules be checked by the researcher for applicability. 1.1. A theoretical framework By placing pattern-based approaches in a lexical seman- tic framework, such as Pustejovsky's Generative Lexicon theory ([16]), my aim is to provide a basis for pattern- based understanding systems …

Content truncated to 500 words