Thai Q-Cor: Integrating word approximation and soundex for Thai query correction

Nowadays, Internet is widely used almost all over the world including Thailand. People can find Web site or information that they need by using search engines like Sansarn or Google. When users type words or phrases into the search box, sometimes they are not satisfied with the returned results. One of the most important problems is misspelled query due to typographical and cognitive errors. To address these errors, we propose Thai query correction system called Thai Q-Cor that is able to verify an inaccurate query and correct it. Our system is composed of two correction modules: word approximation and Soundex. Word approximation module is used for resolving typographical errors by using beam search technique. A userpsilas query will be calculated for error scores by comparing with the words in the dictionary. The words that have the minimal error score will be returned. Soundex module is used for fixing cognitive error by using phoneme search. The userpsilas query must be first converted to phoneme format. The words in the dictionary which have the same phoneme format as the userpsilas query will be returned. Our preliminary result demonstrates the potential of Thai Q-Cor system for repairing the inaccurate userpsilas queries.