A Progressive Algorithm for Cross-Language Information Retrieval Based on Dictionary Translation

Query translation is the mainstream in cross-language information retrieval, but ambiguity must be resolved by methods based on dictionary translation. In this paper, we propose a progressive algorithm for disambiguation which is derived from another algorithm we propose called the max-sum model. The new algorithm take a strategy called weighted-average probability distribution to redistribute the probabilities. Moreover, the new algorithm can be computed in a more direct way by solving an equation system. All the resource our method requires is a bilingual dictionary and a monolingual corpus. Experiments show it outperforms four other methods.