Term-weighting approaches in automatic text retrieval
Information Processing and Management: an International Journal
Information retrieval using a singular value decomposition model of latent semantic structure
SIGIR '88 Proceedings of the 11th annual international ACM SIGIR conference on Research and development in information retrieval
Latent semantic linking over homogeneous repositories
DocEng '01 Proceedings of the 2001 ACM Symposium on Document engineering
An infrastructure for open latent semantic linking
Proceedings of the thirteenth ACM conference on Hypertext and hypermedia
An open linking service supporting the authoring of web documents
Proceedings of the 2002 ACM symposium on Document engineering
Evaluation of video news classification techniques for automatic content personalisation
International Journal of Advanced Media and Communication
Video news classification for automatic content personalization: a genetic algorithm based approach
Proceedings of the 14th Brazilian Symposium on Multimedia and the Web
Hi-index | 0.00 |
The aim of Cross-Language Information Retrieval (CLIR) area is to address situations where a query is made in one language and the application is able to return documents in another. Many CLIR techniques attempt to translate the user's query to the language of the target documents using translation dictionaries. However, these techniques have limitations in terms of lexical coverage of the dictionary adopted. For some applications, the dictionaries are manually edited towards improving the results — but this may require much effort to represent a large collection of information. In this article we propose an infrastructure for defining automatically relationships between Web documents written in different languages. Our approach is based on the Latent Semantic Indexing Technique, which tries to overcome the problems common to the lexical approach due to words with multiple meanings and multiple words with the same meaning. LSI automatically organizes text objects into a semantic structure appropriate for matching [3]. To support the identification of relationships among documents in different languages, the proposed infrastructure manipulates the stem portion of each word in order to index the corresponding Web documents when building the information space manipulated by LSI. To experiment this proposal, we studied the creation of links among news documents in English and Spanish in three different categories: entertainment, technology and world. The results were positive.