Patent ReferencesIndexing subject-locating method System for processing natural language including identifying grammatical rule and semantic concept of an undefined word Patent #: 5109509 InventorsApplicationNo. 596283 filed on 10/11/1990US Classes:704/10, Dictionary building, modification, or prioritization704/9Natural languageExaminersPrimary: Envall, Roy N. Jr.Assistant: Poinvil, Frantzy Attorney, Agent or FirmForeign Patent References
International ClassesG06F 015/38G06F 001/00 Foreign Application Priority Data1989-10-16 JPAbstractA system for creating an index of textual data stores textual data in memory, and a text analyzing module analyzes the textual data and divides it into a plurality of meaningful strings of characters, punctuation marks, symbols, control codes, etc. A dictionary stores sets of specialized words particular to a field of knowledge related to the textual data in a particular language. An entry selecting module selects as index entries only those strings which match one of those specialized words and notes the location(s) of each occurrence each index entry in the text. A printer outputs the selected index entries together with their occurrence positions. Each entry of the dictionary in the specialized field includes information concerning inflections and variants of that entry. The index is quickly and accurately generated by selecting index entries using a specialized dictionary relevant to a particular, specialized field. Since the selection of index entries is made by referring to such a dictionary, differences in criteria for selection used by different operators can be prevented. Since a specialized dictionary is prepared and updated for each field, the knowledge for generating an index is collected and shared by all the operators.Other References
| |