Inducing Translation Lexicons via Diverse Similarity Measures and Bridge Languages

This paper presents a method for inducing translation lexicons between two distant languages without the need for either parallel bilingual corpora or a direct bilingual seed dictionary. The algorithm successfully combines temporal occurrence similarity across dates in news corpora, wide and local cross-language context similarity, weighted Levenshtein distance, relative frequency and burstiness similarity measures. These similarity measures are integrated with the bridge language concept under a robust method of classifier combination for both the Slavic and Northern Indian language families.