No label defined (Q4692986)
From MaRDI portal
!
This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:
scientific article; zbMATH DE number 195029
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | No label defined |
scientific article; zbMATH DE number 195029 |
Statements
5 June 1993
0 references
information retrieval system
0 references
machine learning
0 references
automatic thesaurus generation
0 references
artificial intelligence
0 references
concept learning
0 references
knowledge acquisition
0 references
knowledge verification
0 references
0.7534818649291992
0 references
Starting with an introduction into basic notions of artificial intelligence and expert systems the authors consider possible relationships of these emerging knowledge based technologies to the field of information systems in general and information retrieval in particular. The major part of the book is devoted to an application of AI methods in the area of information retrieval systems, viz. a system with concept learning capabilities in order to support searching large bibliographic databases. That system is part of TUBIBMUE, a routinely used bibliographic in-house information retrieval and library system developed over the years at the Technische Universität München. After a description of its architecture and main functionality the authors turn to TUBIBMUE's knowledge based extension in terms of the concept learning component TEGEN (which stands for thesaurus generation). The foundations of the learning procedure, system architecture as well as implementation strategies for TEGEN are dealt with in-depth. TEGEN is based on a technique called ``learning by analyzing'' which is a refinement of the well-known ``learning by observation'' approach. The system is able to infer conceptual relationships among search terms based on the syntactic structure of query formulations of a conventional boolean query language that also includes truncation and adjacency operators. The relationships TEGEN identifies on the basis of observing the users' behavior during online retrieval dialogs are tentative and must be confirmed by later observation data in order to become part of the thesaurus that is used as a background knowledge source for the automatic enhancement of search queries. TEGEN consists of three major parts: a thesaurus that incrementally grows as the learning process of the system proceeds, a knowledge acquisition and a knowledge verification module actually controlling the learning process. The thesaurus contains a list of keywords which are linked by common thesaurus relations plus some search-specific extensions such as positive, negative and candidate term lists as well as a special root dictionary of search terms. The knowledge acquisition component produces the intermediate results from the observation of user queries based on a set of knowledge acquisition rules that either are capable of automatically deducing conceptual relationships among terms or, alternatively, cause the system to ask users by way of immediate feedback for interactive clarification of, e.g., uncertain or ambiguous conceptual relationships. The knowledge verification component, finally, is equally based on a set of verification rules and is activated when transferring the intermediate results from the knowledge acquisition phase into final results leading to a thesaurus update. As with the initial knowledge acquisition process the knowledge verification procedure is not entirely automated, but interactively takes into account the expertise of different users by means of explicit feedback processes. TEGEN's learning capabilities are encoded in a set of production rules that are executed in a forward chaining mode. It is implemented in PASCAL thus - following the claims of the authors - realizing the paradigm of knowledge based programming while at the same time easing its integration into conventional programming and (database) system environments.
0 references