Językowe problemy optymalizacji wyszukiwania informacji w systemie iSybislaw
Linguistic Problems of Information Retrieval Optimization in iSybislaw System
Languages of publication
During the past dozen or so years there have occurred large changes in the information society and computer technologies. An increase in the information needs of the users of information retrieval systems (IRS) has been observed. The methodology of information processing has changed: user unfriendly and complicated information retrieval languages (IRL) have been abandoned and the use of natural languages in their metalinguistic function has been reestablished with the use of keywords being its symptom. In the commonly used IRS the structure of the keyword language is flat, the keywords are nonsubordinate. The keyword language, used in iSybislaw (in the bibliographical information retrieval system of the world Slavic linguistics to be more exact), is non-standard in its character. It differs from other keyword languages with its paradigmatic semantic structure. The methodology of processing multilingual vocabulary takes into consideration the relations of equivalence and polysemy. Establishing the equivalence classes (consisting of terminological equivalents and other type equivalents including: personal, geographic and corporative names among others) in an IRS allows the retrieval of relevant information by means of any given search term: in any language and alphabet. A certain type of preference will be introduced in the case of equivalence in the form of descriptors (and non-descriptors), whereas in the case of polysemy a system of (meta)metalinguistic appositions will be used. The use of a descriptor system in an IRS such as iSybislaw ensures the completeness of retrieved information, resolving the issue of polysemy on the other hand allows for a more effective information retrieval measured by the precision and relevancy of retrieved data.
Publication order reference