Probabilistic Approach to Crosslingual Information Retrieval

摘要

We present a method to translate queries from an arbitrary source language to retrieve documents in a destination language merely with easily obtainable instruments such as a machine readable dictionary and monolingual corpora in both languages. The key is to infer probabilistical information about the query and structuring the destination language terms accordingly. Though the results compare unfavourably with those obtained with more sophisticated but difficult to obtain IR-methods using Part-of-Speech-Tagging and/or Phrase dictionaries, our work shows the successful deployment and combination of related work to crosslingual Information Retrieval.