Digital Library[ Search Result ]
KMSS: Korean Media Script Dataset for Dialogue Summarization
Bong-Su Kim, Ji-Yoon Kim, Seung-ho Choi, Hyun-Kyu Jeon, Hye-Jin Jun, Hye-In Jung, Jung-Hoon Jang
http://doi.org/10.5626/JOK.2024.51.4.311
Dialogue summarization involves extracting or generating key contents from multi-turn documents consisting of utterances by multiple speakers. Dialogue summarization models are beneficial in analyzing content and service records for recommendations in conversation systems. However, there are no Korean dialogue summarization datasets necessary for model construction. This paper proposes a dataset for generative-based dialogue summarization. Source data were collected from the large-capacity contents of domestic broadcasters, and annotators manually labeled them. The dataset comprises approximately 100,000 entries across 6 categories, with summary sentences annotated as single sentences, three sentences, or two-and-a-half sentences. Additionally, this paper introduces a dialogue summary labeling guide to internalize and control data characteristics. It also presents a method for selecting a decoding model structure for model suitability verification. Through experiments, we highlight some characteristics of the constructed data and present benchmark performances for future research.
Query-based Abstractive Summarization Model Using Sentence Ranking Scores and Graph Techniques
http://doi.org/10.5626/JOK.2020.47.12.1172
The purpose of the fundamental abstractive summarization model is to generate a short summary document that includes all important contents within the document. Conversely, in the query-based abstractive summarization model, information related to the query should be selected and summarized within the document. The existing query-based summarization models calculates the importance of sentences using only the weight of words through an attention mechanism between words in the document and the query. This method has a disadvantage in that it is difficult to reflect the entire context information of the document to generate an abstractive summary. In this paper, we resolve this problems by calculating the sentence ranking scores and a sentence-level graph structure. Our proposed model shows higher performance than the previous research model, 1.44%p in ROUGE-1 and 0.52%p in ROUGE-L.
Search
Journal of KIISE
- ISSN : 2383-630X(Print)
- ISSN : 2383-6296(Electronic)
- KCI Accredited Journal
Editorial Office
- Tel. +82-2-588-9240
- Fax. +82-2-521-1352
- E-mail. chwoo@kiise.or.kr