Mongolian Questions Classification in the Law Domain

2020 
Question classification is an important part of the question answering system. Through question classification, we can understand the purpose of users’ questions and determine the conditions that the answers need to meet. This paper introduces our work on Mongolian questions classification in the law domain. Mongolian is a kind of low-resource language, and there is a lack of public legal-oriented corpus. The complicated morphological structure in Mongolian vocabulary causes the data-sparse problem. In this paper, we constructed a legaloriented corpus for the Mongolian questions classification. And we proposed a method of Mongolian questions classification based the combination of the BERT and BiLSTM model. The BERT model is proposed to extract the semantic features representations of the text. And then the words semantic features acquired are input into the BiLSTM model for questions classification. Our results on the experiments confirm that the method proposed in this paper works effectively with the F 1 value of 86.98%.
    • Correction
    • Source
    • Cite
    • Save
    • Machine Reading By IdeaReader
    18
    References
    0
    Citations
    NaN
    KQI
    []