A Method of Automatic Hypertext Construction from an Encyclopedic Dictionary of a Specific Field
A Method of Automatic Hypertext Construction from an Encyclopedic Dictionary of a Specific Field
复制标题
专业百科词典自动构造超文本的方法
DOI：
10.3115/974499.974546
复制
发表时间：
1992
期刊：
影响因子：
--
通讯作者：
Masahiko Murakami
中科院分区：
文献类型：
--
作者：
S. Kurohashi;M. Nagao;Satoshi Sato;Masahiko Murakami
1 Introduction Nowadays, very large volume of texts are created and stored in computer, and as a result the retrieval of texts which fits to a user's demand has become a difficult problem. Hypertext is a typical system to answer this problem , whose primary objective is to establish flexible as-sociative links between relevant text parts and to allow users to select and trace links to see relevant text contents which are connected by links. A difficult problem here is how to construct automatically a network structure in a given set of text data. This paper is concerned with (1) automatic conversion of a plain text set into a hypertext structure, and (2) construction of flexible human interface for the hypertext system. We applied natural language processing methods to locate important conceptual terms in a text corpus and to establish varieties of links between these terms and appropriate text portions. 2 Extraction of thesaurus information The text corpus we handled as a concrete example was the Encyclopedic Dictionary of Computer Science (hereafter abbreviated as EDCS. Iwanami Publ. 1990. En-glish translation will appear soon from Academic Press). It includes 4500 terms and has the text volume of two million Japanese characters (4 Mega bytes). The first part of the term description of EDCS is devoted to synonyms, antonyms, abbreviations and broader concept words. This part has typical sentential styles such as, (ii) A is abbreviated as B. (iv) A stands for B. (v) We call A S {for short}. By finding these sentential patterns the relation between the words A and B is established as follows. (i) p-link is set up from a synonym word to a sentence which defines the synonym relation. (ii) s-link (by synonym) is set up from a defined word to defining words by synonym relation. Typical sentential styles of intensional definition are: (i) A is defined as B. A is regarded as B. (ii) A means B. A connotes B. A is B. By identifying these patterns in a term description part, the relation between the defined word (A) and the definition sentences is established as: (i) p-link is set up from the defined word to the definition sentence when the defined word is not the headword of the term description. This is the case when the defined word is not so important as a headword of the dictionary, and so a rather simple definition description …
