Attitude detection for one-round conversation: Jointly extracting target-polarity pairs

Zhaohao Zeng, Pingping Lin, Ruihua Song, Tetsuya Sakai

    Research output: Chapter in Book/Report/Conference proceedingConference contribution

    Abstract

    We tackle Attitude Detection, which we define as the task of extracting the replier's attitude, i.e., a target-polarity pair, from a given one-round conversation. While previous studies considered Target Extraction and Polarity Classification separately, we regard them as subtasks of Attitude Detection. Our experimental results show that treating the two subtasks independently is not the optimal solution for Attitude Detection, as achieving high performance in each subtask is not sufficient for obtaining correct target-polarity pairs. Our jointly trained model AD-NET substantially outperforms the separately trained models by alleviating the target-polarity mismatch problem. Moreover, we proposed a method utilising the attitude detection model to improve retrieval-based chatbots by re-ranking the response candidates with attitude features. Human evaluation indicates that with attitude detection integrated, the new responses to the sampled queries from are statistically significantly more consistent, coherent, engaging and informative than the original ones obtained from a commercial chatbot.

    Original languageEnglish
    Title of host publicationWSDM 2019 - Proceedings of the 12th ACM International Conference on Web Search and Data Mining
    PublisherAssociation for Computing Machinery, Inc
    Pages285-293
    Number of pages9
    ISBN (Electronic)9781450359405
    DOIs
    Publication statusPublished - 2019 Jan 30
    Event12th ACM International Conference on Web Search and Data Mining, WSDM 2019 - Melbourne, Australia
    Duration: 2019 Feb 112019 Feb 15

    Publication series

    NameWSDM 2019 - Proceedings of the 12th ACM International Conference on Web Search and Data Mining

    Conference

    Conference12th ACM International Conference on Web Search and Data Mining, WSDM 2019
    CountryAustralia
    CityMelbourne
    Period19/2/1119/2/15

    Keywords

    • Attitude detection
    • Chatbot
    • Conversation
    • Sentiment analysis

    ASJC Scopus subject areas

    • Computer Networks and Communications
    • Software
    • Computer Science Applications

    Cite this

    Zeng, Z., Lin, P., Song, R., & Sakai, T. (2019). Attitude detection for one-round conversation: Jointly extracting target-polarity pairs. In WSDM 2019 - Proceedings of the 12th ACM International Conference on Web Search and Data Mining (pp. 285-293). (WSDM 2019 - Proceedings of the 12th ACM International Conference on Web Search and Data Mining). Association for Computing Machinery, Inc. https://doi.org/10.1145/3289600.3291038

    Attitude detection for one-round conversation : Jointly extracting target-polarity pairs. / Zeng, Zhaohao; Lin, Pingping; Song, Ruihua; Sakai, Tetsuya.

    WSDM 2019 - Proceedings of the 12th ACM International Conference on Web Search and Data Mining. Association for Computing Machinery, Inc, 2019. p. 285-293 (WSDM 2019 - Proceedings of the 12th ACM International Conference on Web Search and Data Mining).

    Research output: Chapter in Book/Report/Conference proceedingConference contribution

    Zeng, Z, Lin, P, Song, R & Sakai, T 2019, Attitude detection for one-round conversation: Jointly extracting target-polarity pairs. in WSDM 2019 - Proceedings of the 12th ACM International Conference on Web Search and Data Mining. WSDM 2019 - Proceedings of the 12th ACM International Conference on Web Search and Data Mining, Association for Computing Machinery, Inc, pp. 285-293, 12th ACM International Conference on Web Search and Data Mining, WSDM 2019, Melbourne, Australia, 19/2/11. https://doi.org/10.1145/3289600.3291038
    Zeng Z, Lin P, Song R, Sakai T. Attitude detection for one-round conversation: Jointly extracting target-polarity pairs. In WSDM 2019 - Proceedings of the 12th ACM International Conference on Web Search and Data Mining. Association for Computing Machinery, Inc. 2019. p. 285-293. (WSDM 2019 - Proceedings of the 12th ACM International Conference on Web Search and Data Mining). https://doi.org/10.1145/3289600.3291038
    Zeng, Zhaohao ; Lin, Pingping ; Song, Ruihua ; Sakai, Tetsuya. / Attitude detection for one-round conversation : Jointly extracting target-polarity pairs. WSDM 2019 - Proceedings of the 12th ACM International Conference on Web Search and Data Mining. Association for Computing Machinery, Inc, 2019. pp. 285-293 (WSDM 2019 - Proceedings of the 12th ACM International Conference on Web Search and Data Mining).
    @inproceedings{34cd61473a764e9384495b1270f59927,
    title = "Attitude detection for one-round conversation: Jointly extracting target-polarity pairs",
    abstract = "We tackle Attitude Detection, which we define as the task of extracting the replier's attitude, i.e., a target-polarity pair, from a given one-round conversation. While previous studies considered Target Extraction and Polarity Classification separately, we regard them as subtasks of Attitude Detection. Our experimental results show that treating the two subtasks independently is not the optimal solution for Attitude Detection, as achieving high performance in each subtask is not sufficient for obtaining correct target-polarity pairs. Our jointly trained model AD-NET substantially outperforms the separately trained models by alleviating the target-polarity mismatch problem. Moreover, we proposed a method utilising the attitude detection model to improve retrieval-based chatbots by re-ranking the response candidates with attitude features. Human evaluation indicates that with attitude detection integrated, the new responses to the sampled queries from are statistically significantly more consistent, coherent, engaging and informative than the original ones obtained from a commercial chatbot.",
    keywords = "Attitude detection, Chatbot, Conversation, Sentiment analysis",
    author = "Zhaohao Zeng and Pingping Lin and Ruihua Song and Tetsuya Sakai",
    year = "2019",
    month = "1",
    day = "30",
    doi = "10.1145/3289600.3291038",
    language = "English",
    series = "WSDM 2019 - Proceedings of the 12th ACM International Conference on Web Search and Data Mining",
    publisher = "Association for Computing Machinery, Inc",
    pages = "285--293",
    booktitle = "WSDM 2019 - Proceedings of the 12th ACM International Conference on Web Search and Data Mining",

    }

    TY - GEN

    T1 - Attitude detection for one-round conversation

    T2 - Jointly extracting target-polarity pairs

    AU - Zeng, Zhaohao

    AU - Lin, Pingping

    AU - Song, Ruihua

    AU - Sakai, Tetsuya

    PY - 2019/1/30

    Y1 - 2019/1/30

    N2 - We tackle Attitude Detection, which we define as the task of extracting the replier's attitude, i.e., a target-polarity pair, from a given one-round conversation. While previous studies considered Target Extraction and Polarity Classification separately, we regard them as subtasks of Attitude Detection. Our experimental results show that treating the two subtasks independently is not the optimal solution for Attitude Detection, as achieving high performance in each subtask is not sufficient for obtaining correct target-polarity pairs. Our jointly trained model AD-NET substantially outperforms the separately trained models by alleviating the target-polarity mismatch problem. Moreover, we proposed a method utilising the attitude detection model to improve retrieval-based chatbots by re-ranking the response candidates with attitude features. Human evaluation indicates that with attitude detection integrated, the new responses to the sampled queries from are statistically significantly more consistent, coherent, engaging and informative than the original ones obtained from a commercial chatbot.

    AB - We tackle Attitude Detection, which we define as the task of extracting the replier's attitude, i.e., a target-polarity pair, from a given one-round conversation. While previous studies considered Target Extraction and Polarity Classification separately, we regard them as subtasks of Attitude Detection. Our experimental results show that treating the two subtasks independently is not the optimal solution for Attitude Detection, as achieving high performance in each subtask is not sufficient for obtaining correct target-polarity pairs. Our jointly trained model AD-NET substantially outperforms the separately trained models by alleviating the target-polarity mismatch problem. Moreover, we proposed a method utilising the attitude detection model to improve retrieval-based chatbots by re-ranking the response candidates with attitude features. Human evaluation indicates that with attitude detection integrated, the new responses to the sampled queries from are statistically significantly more consistent, coherent, engaging and informative than the original ones obtained from a commercial chatbot.

    KW - Attitude detection

    KW - Chatbot

    KW - Conversation

    KW - Sentiment analysis

    UR - http://www.scopus.com/inward/record.url?scp=85061744733&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=85061744733&partnerID=8YFLogxK

    U2 - 10.1145/3289600.3291038

    DO - 10.1145/3289600.3291038

    M3 - Conference contribution

    AN - SCOPUS:85061744733

    T3 - WSDM 2019 - Proceedings of the 12th ACM International Conference on Web Search and Data Mining

    SP - 285

    EP - 293

    BT - WSDM 2019 - Proceedings of the 12th ACM International Conference on Web Search and Data Mining

    PB - Association for Computing Machinery, Inc

    ER -