•   Quick Search:       Advanced Search
    LI Shengwang,YANG Yi,XU Yunfeng,ZHANG Yan.A survey of text aspect-based sentiment classification[J].Journal of Hebei University of Science and Technology,2020,41(6):518-527
    文本方面級情感分類方法綜述
    A survey of text aspect-based sentiment classification
    Received:October 02, 2020  Revised:November 06, 2020
    DOI:10.7535/hbkd.2020yx06006
    中文關鍵詞:  自然語言處理  情感分類  方面級別  文本分類  深度學習  圖神經網絡  圖卷積網絡
    英文關鍵詞:natural language processing  sentiment classification  aspect-based  text classification  deep learning  graph neural network  graph convolutional network
    基金項目:中國留學基金委地方合作項目(201808130283); 中國教育部人工智能協同育人項目(201801003011); 河北科技大學校立課題(82/1182108)
    Author NameAffiliationE-mail
    LI Shengwang School of Information Science and Engineering, Hebei University of Science and Technology, Shijiazhuang  
    YANG Yi School of Information Science and Engineering, Hebei University of Science and Technology, Shijiazhuang  
    XU Yunfeng School of Information Science and Engineering, Hebei University of Science and Technology, Shijiazhuang hbkd_xyf@hebust.edu.cn 
    ZHANG Yan School of Information Science and Engineering, Hebei University of Science and Technology, Shijiazhuang  
    Hits: 1577
    Download times: 1742
    中文摘要:
          隨著深度學習的發展,方面級情感分類已經在單領域和單一語言中取得了大量的研究成果,但是在多領域的研究還有提升的空間。通過對近年來文本方面級情感分類方法進行歸納總結,介紹了情感分類的具體應用場景,整理了方面級情感分類常用的數據集,并對方面級情感分類的發展進行了總結與展望,提出未來可在以下領域開展深入研究:1)探索基于圖神經網絡的方法,彌補深度學習方法存在的局限性;2)學習融合多模態數據,豐富單一文本的情感信息;3)開展更多針對多語言文本和低資源語言的研究。
    英文摘要:
          With the development of deep learning, aspect-based sentiment classification has achieved a lot of results in a single field and a single language, but there is room for improvement in multi-fields. By summarizing up the methods of text aspect-based sentiment classification in recent years, the specific application scenarios of sentiment classification were introduced, and the commonly used data sets of aspect-based sentiment classification were categorized. The development of aspect-based sentiment classification were summarized and prospected, and further research can be carried out in the following areas: exploring methods based on graph neural networks to make up for the limitations of deep learning methods; learning to fuse multi-modal data to enrich the emotional information of a single text; developing more targeted research work on multilingual texts and low-resource languages.
    View Full Text  View/Add Comment  Download reader
    Close
    购乐彩