Matching Thai Profession Descriptions from Different Sources Using English as Interlingual Language

This paper proposes a method to improve text matching for occupation descriptions from different standards in Thailand. To solve ambiguity from technical compound words from standards, we apply machine translation, and similarity calculation for matching occupation and profession from sources that focused on different aspects. Thus, despite mentioning on same occupation, the descriptions from different standards share very little terms for matching the job across standards. By translating terms from Thai to English, we solve the issue of synonym technical terms varily used in different sources to improve similarity scores. Furthermore, tree structure of the standards is considered to assist on limiting search space. From experiments, the results of the baseline is very low, and the proposed method averagely exceed the result from the baseline for more than 6 percents for top-3 matching and 5 percents for top-5 matching in terms of accuracy. The use of both Thai and English translated job descriptions for matching using similar terms can noticeably increase a capability of matching jobs that cannot be found using only original Thai job description by adding more features of English-translated terms.