Who Speaks Like a Style of Vitamin: Towards Syntax-Aware Dialogue Summarization Using Multi-Task Learning
Abstractive dialogue summarization is a challenging task for several reasons. First, most of the key information in a conversation is scattered across utterances through multi-party interactions with different textual styles. Second, dialogues are often informal structures, wherein different individ...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2021-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/9664379/ |
| _version_ | 1828144599113138176 |
|---|---|
| author | Seolhwa Lee Kisu Yang Chanjun Park Joao Sedoc Heuiseok Lim |
| author_facet | Seolhwa Lee Kisu Yang Chanjun Park Joao Sedoc Heuiseok Lim |
| author_sort | Seolhwa Lee |
| collection | DOAJ |
| description | Abstractive dialogue summarization is a challenging task for several reasons. First, most of the key information in a conversation is scattered across utterances through multi-party interactions with different textual styles. Second, dialogues are often informal structures, wherein different individuals express personal perspectives, unlike text summarization, tasks that usually target formal documents such as news articles. To address these issues, we focused on the association between utterances from individual speakers and unique syntactic structures. Speakers have unique textual styles that can contain linguistic information, such as voiceprint. To do this, we used ad-hoc analysis to explore speakers’ text styles and constructed a syntax-aware model by leveraging linguistic information (i.e., POS tagging), which alleviates the above issues by inherently distinguishing utterances from individual speakers. Our approach allows for both data and model-centric investigation. Also, we employed multi-task learning of both syntax-aware information and dialogue summarization. To the best of our knowledge, our approach is the first method to apply multi-task learning to the dialogue summarization task. Experiments on a SAMSum corpus (a large-scale dialogue summarization corpus) demonstrated that our method improved upon the vanilla model. Consequently, we found that our efforts of syntax-aware approach have been reflected by the model. |
| first_indexed | 2024-04-11T20:18:54Z |
| format | Article |
| id | doaj.art-223ffc4b844c409380dab66484c152ce |
| institution | Directory Open Access Journal |
| issn | 2169-3536 |
| language | English |
| last_indexed | 2024-04-11T20:18:54Z |
| publishDate | 2021-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj.art-223ffc4b844c409380dab66484c152ce2022-12-22T04:04:53ZengIEEEIEEE Access2169-35362021-01-01916888916889810.1109/ACCESS.2021.31245569664379Who Speaks Like a Style of Vitamin: Towards Syntax-Aware Dialogue Summarization Using Multi-Task LearningSeolhwa Lee0https://orcid.org/0000-0002-8109-8497Kisu Yang1https://orcid.org/0000-0003-4983-0307Chanjun Park2https://orcid.org/0000-0002-7200-9632Joao Sedoc3Heuiseok Lim4https://orcid.org/0000-0002-9269-1157Department of Computer Science, University of Copenhagen, Copenhagen, DenmarkVAIV Corporation, Seoul, Republic of KoreaDepartment of Computer Science and Engineering, Korea University, Seoul, Republic of KoreaDepartment of Technology, Operations and Statistics, New York University, New York, NY, USADepartment of Computer Science and Engineering, Korea University, Seoul, Republic of KoreaAbstractive dialogue summarization is a challenging task for several reasons. First, most of the key information in a conversation is scattered across utterances through multi-party interactions with different textual styles. Second, dialogues are often informal structures, wherein different individuals express personal perspectives, unlike text summarization, tasks that usually target formal documents such as news articles. To address these issues, we focused on the association between utterances from individual speakers and unique syntactic structures. Speakers have unique textual styles that can contain linguistic information, such as voiceprint. To do this, we used ad-hoc analysis to explore speakers’ text styles and constructed a syntax-aware model by leveraging linguistic information (i.e., POS tagging), which alleviates the above issues by inherently distinguishing utterances from individual speakers. Our approach allows for both data and model-centric investigation. Also, we employed multi-task learning of both syntax-aware information and dialogue summarization. To the best of our knowledge, our approach is the first method to apply multi-task learning to the dialogue summarization task. Experiments on a SAMSum corpus (a large-scale dialogue summarization corpus) demonstrated that our method improved upon the vanilla model. Consequently, we found that our efforts of syntax-aware approach have been reflected by the model.https://ieeexplore.ieee.org/document/9664379/Summarizationdialogue summarizationmulti-task learningsyntax-aware conversationdeep learning |
| spellingShingle | Seolhwa Lee Kisu Yang Chanjun Park Joao Sedoc Heuiseok Lim Who Speaks Like a Style of Vitamin: Towards Syntax-Aware Dialogue Summarization Using Multi-Task Learning IEEE Access Summarization dialogue summarization multi-task learning syntax-aware conversation deep learning |
| title | Who Speaks Like a Style of Vitamin: Towards Syntax-Aware Dialogue Summarization Using Multi-Task Learning |
| title_full | Who Speaks Like a Style of Vitamin: Towards Syntax-Aware Dialogue Summarization Using Multi-Task Learning |
| title_fullStr | Who Speaks Like a Style of Vitamin: Towards Syntax-Aware Dialogue Summarization Using Multi-Task Learning |
| title_full_unstemmed | Who Speaks Like a Style of Vitamin: Towards Syntax-Aware Dialogue Summarization Using Multi-Task Learning |
| title_short | Who Speaks Like a Style of Vitamin: Towards Syntax-Aware Dialogue Summarization Using Multi-Task Learning |
| title_sort | who speaks like a style of vitamin towards syntax aware dialogue summarization using multi task learning |
| topic | Summarization dialogue summarization multi-task learning syntax-aware conversation deep learning |
| url | https://ieeexplore.ieee.org/document/9664379/ |
| work_keys_str_mv | AT seolhwalee whospeakslikeastyleofvitamintowardssyntaxawaredialoguesummarizationusingmultitasklearning AT kisuyang whospeakslikeastyleofvitamintowardssyntaxawaredialoguesummarizationusingmultitasklearning AT chanjunpark whospeakslikeastyleofvitamintowardssyntaxawaredialoguesummarizationusingmultitasklearning AT joaosedoc whospeakslikeastyleofvitamintowardssyntaxawaredialoguesummarizationusingmultitasklearning AT heuiseoklim whospeakslikeastyleofvitamintowardssyntaxawaredialoguesummarizationusingmultitasklearning |