A morpho-syntactic categorial modeling for Korean is introduced. Variable categories and notations for word order treatment are newly invented and notations for elliptic arguments are also suggested. Incremental parsing technique of a morpheme graph is newly developed using the proposed categorial Korean modeling. Two heuristics based on the coverage of sub-trees and the part-of-speech bigrams are newly developed to select the best parse from ambiguous parses. 1 Introduction Korean is a non-configurational, postpositional and agglutinative language. Postpositions such as noun-endings, verb-endings, and prefinal verbendings are morphemes which represent NP's and VP's role in a sentence or transform VP's into NP's/AP's. Since a sequence of prefinal verb-endings, auxiliary verbs and verb-endings can make hundreds of different usages of verbs, morpheme-based grammar modeling is considered as a natural consequence for Korean. In this paper, we tried to model Korean through a directional ca...

An internet search in a multi-agent environment can most naturally be defined as an interactive query refinment process where users can modify previous queries to satisfy an information need. In this paper, we present an NLI agent that supports KQML-based agent communication and interactive query refinement using natural language discourse processing technique. The NLI agent supports natural language queries and commands along with a search history so that users can refine their queries based on previous search results. Users communicate with the NLI agent using natural language and the web search agent receives users' requests through compiled KQML/KIF messages. The web search agent automatically generates standard boolean queries that can interface with many currently available standard web search engines. The NLI agent employs natural language semantic and discourse processing techniques enabling users to use anaphoric terms and elided expressions freely. Currently, large amount of ...

Both dictionary-based and rule-based methods on grapheme-to-phoneme conversion have their own advantages and limitations. For example, a large sized phonetic dictionary and complex morphophonemic rules are required for the dictionary-based method and the LTS(letter to sound) rule-based method itself cannot model the complete morphophonemic constraints. This paper describes a grapheme-to-phoneme conversion method using a dictionarybased and rule-based hybrid method with a phonetic pattern dictionary and CCV (consonant consonant vowel) LTS (letter to sound) rules. The phonetic pattern dictionary, standing for the dictionary-based method, contains entries in the form of a morpheme pattern and its phonetic pattern. The patterns represent candidate phonological changes in left and right boundaries of morphemes. Obviously, the CCV LTS rules stands for the rule-based method. The rules are in charge of grapheme-to-phoneme conversion within morphemes. The conversion method consists of mainly tw...

Internet searchinamulti-agentenvironment can most naturally be modeled as an interactive query refinment process where users can refine and modify their previous queries to satisfy their information need. In this paper, we present a NLI agent that supports kqml-based agent communication and the interactive query refinement using natural language discourse processing technique. The NLI agent supports natural language queries and commands along with a search history so that users can refine their queries based on the previous search results. Users communicate with the NLI agent using natural language and the web searchagent receives user's requests through compiled kqml/kif messages. The web searchagent automatically generates standard boolean queries that can interface with manycurrently available standard web search engines. The NLI agent employs natural language semantic and discourse processing techniques enabling users to use anaphoric terms and elided expressions freely. Currently, large amount of query and command corpus are being collected to tune the natural language and discourse processing in the NLI agent. Key words: query refinement, natural language processing, web search, multiagent, discourse processing This research was partly supported by Ministry of Science and Technology Soft Science Project and KOSEF (teuk-ceong-ki-cho) no.970-1020-3 (1997). 1 1

This paper describes a grapheme-to-phoneme conversion method using phoneme connectivity and CCV conversion rules. The method consists of mainly four modules including morpheme normalization, phrase-break detection, morpheme to phoneme conversion and phoneme connectivity check.