A three-dimensional model of semantic search: queries, resources, and results

We propose three-dimensional model of semantic search that analyzes search requests, information resources (IRs) and search results. This model is proposed as an additional tool for describing and comparing information retrieval systems (IRSs) that use various elements of artificial intelligence and knowledge management for more effective and relevant satisfaction of user information needs. In this work we analyze existing approaches to the semanticization of search queries and the use of external knowledge sources for retrieval process. The values of parameters analyzed by this model are not mutually exclusive, that is, the same IRS can support several search options. More over, the representation means of queries and resources are not always comparable. The model makes it possible to identify IRSs with intersected triads «request-IR-result» and to perform their comparison precisely on these subclasses of search problems. This approach allows to select search algorithms that are more pertinent for specific user tasks and to choose on base of this selection appropriate retrieval services that provide information for further processing. An important feature of the proposed model is that it uses only those IRS characteristics that can be directly evaluated by retrieval users.