Application of Multilingual Thesauri for Cross Language Information Retrieval (CLIR): Amharic • English Cross Language Information Retrieval for Legal Environment
No Thumbnail Available
Date
2005-06
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Addis Ababa University
Abstract
It is crucial for cross language retrieval tools to be capable of translating queries
and/or documents to make information accessible. In areas where such efficient
systems are lacking for language pairs like Amharic and English, intermediate tools
like the thesauri can be used to make translation of queries possible.
This research has developed parallel thesauri for business law to be used in testing
the retrieval of thesauri based translation of queries across the two languages. The
thesauri are developed by taking the commercial code of Ethiopia as a
representative corpus and domain experts in law were involved to make the
conceptual analysis and facet determination. This procedure was supplemented by
machine assisted indexing.
An in-house retrieval system has been developed to test the application of the
multilingual thesauri developed for this purpose. Thesauri based retrieval is tested
where concept based translation of queries is attempted as compared to word based
translations as in the case of machine translation. Queries have been collected from
legal experts and a document collection in the area of law has been developed from
research abstracts for this purpose. To test retrieval performance, queries are
translated to their equivalent concepts using the thesauri and the equivalent
concepts are used to query the collection in the target language.
Retrieval outputs measured in terms of precision and recall show promising results
as they have managed to retrieve relevant documents across languages. However,
the obtained performance can be made better if other resources are made available
as indicated in the recommendation. Therefore, these results have made the
recommendation of other tools as well as approaches as being important to arrive at
better retrieval performance.
Description
Keywords
Information Science