Semantic Measurement of Related Degree between Unknown Word and Related Word for Automatic Extension of Lexical Dictionary

This paper deals with a research of semantic information retrieval, especially is to measure related degree between unknown word (UW), which is a big obstacle in an area of semantic document retrieval, and related word. To do this, we consider UW as word not defined in WordNet that is English lexical dictionary and grasp concepts related with UW from Web documents through co-occurrence and measure related degree using probabilistic and semantic method. This paper describes just a method for the measurement except an objective and quantitative evaluation comparing with other systems or results. However, we show possibility, expectation and value of the research using some kinds of sample documents.