PALKA: A system for lexical knowledge acquisition

Jun Tae Kim, Dan I. Moldovan

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

6 Scopus citations

Abstract

In memory-based natural language processing, semantic phrasal patterns constitute a main knowledge source for recognizing input sentences. Although the memory-based parsing approach has many advantages, one significant problem is that a very large number of semantic phrasal patterns is needed. Manual creation of patterns is unrealistic, and even for a small application domain, it is very time consuming. To solve the scalability problem, automatic acquisition of semantic patterns must be provided. In this paper, we present a practical approach to solve this problem. The PALKA (Parallel Automatic Linguistic Knowledge Acquisition) system acquires semantic patterns from a set of domain specific sample texts and their desired output representations. The acquired patterns are tuned further through generalizations and specializations of semantic constraints. This paper presents the PALKA system and its application to a set of news articles.

Original languageEnglish
Title of host publicationProc 2 Int Conf Inf Knowl Manage
EditorsBharat Bhargava, Timothy Finin, Yelena Yesha
PublisherPubl by ACM
Pages124-131
Number of pages8
ISBN (Print)0897916263
StatePublished - 1993
EventProceedings of the 2nd International Conference on Information and Knowledge Management - Washington, DC, USA
Duration: 1 Nov 19935 Nov 1993

Publication series

NameProc 2 Int Conf Inf Knowl Manage

Conference

ConferenceProceedings of the 2nd International Conference on Information and Knowledge Management
CityWashington, DC, USA
Period1/11/935/11/93

Fingerprint

Dive into the research topics of 'PALKA: A system for lexical knowledge acquisition'. Together they form a unique fingerprint.

Cite this