意味および文脈情報を用いた日本語文の解析 : 文脈を考慮した処理

概要

論文の詳細を見る
In the previous paper, we explained the semantic descriptions of verbs and nouns.In this paper, we explicate how contextual information is organized in our system and how contextual information and semantic descriptions are combined in order to perform contextual analysis, for example, to fill up omitted words from preceding sentences. In most of the artificial intelligence approaches to computer understanding of natural language, logical expressions are used to represent context, and rather complex logical operations are required to perform even a simple contextual analysis. On the contrary we memorize context in the form similar to the short-term memory of human beings. According to several heuristic's the words in the memory are re-ordered in order to emphasize the theme words which express the key subjects of the sentences. We describe the detailed construction of this memory. Several examples of complicated sentences parsed by our system are also shown in this paper.
一般社団法人情報処理学会の論文
1976-01-15