Website Term Browser: Overcoming language barriers in text retrieval

Issue title: IBERAMIA '02

Article type: Research Article

Authors: Peñas, Anselmo | Verdejo, Felisa | Gonzalo, Julio

Affiliations: Departamento de Lenguajes y Sistemas Informáticos, Universidad Nacional de Educación a Distancia (UNED), Ciudad Universitaria, s/n, 28040 Madrid, Spain. E-mail: {anselmo, julio, felisa}@lsi.uned.es; URL: http://nlp.uned.es

Abstract: Current search systems fail to satisfy users when the relevant information is written in a foreign language; when the user is not aware of the relevant – perhaps specialized – terminology for a given topic; or when the user need is fuzzy and requires assisted search once inside an appropriate web portal. This paper describes an interactive multilingual search system that alleviates such limitations, through the browsing of phrases in different languages after being automatically extracted from the text collection. The evaluation of WTB has been focussed in two aspects: the capability to offer translingual terminology to users, and the usefulness of phrase browsing. In this sense, the evaluation shows that users consider the new level of terminological information useful, as it complements the traditional document ranking outcome.

Journal: Journal of Intelligent & Fuzzy Systems, vol. 12, no. 3-4, pp. 221-233, 2002