| 摘要 |
<p>A search technique based on classification of database records in accordance with a predominant concept or feature, to facilitate searching of the database by specifying a concept rather than detailed textual or other content as a basis for searching. Each database record is processed prior to searching, to establish a key concept for the record (step 66, FIG. 3), and the concepts are defined in relation to a thesaurus-like hierarchy of concepts (46). For each database record, a key concept, or its location in the hierarchy is stored (68) in association with the database record, to facilitate later searching of the database using key concepts rather than textual or other specific content. An important aspect of the invention is determination of the key concept location for each database record, by computing a distance value for each node (steps 124, 126 and 128, FIG. 5), from a combination of coverage values (104) and anticoverage values (112), and then selecting (132) the node with the lowest distance value as the desired location of the predominant concept. <IMAGE></p> |
