Identifying Topic Evolutionary Patterns Based on Component Decomposition of Temporal Sequence

• Item type: SIG Technical Reports(1)
• 公開日: 2019-12-16
• タイトル: Identifying Topic Evolutionary Patterns Based on Component Decomposition of Temporal Sequence
• タイトル / 言語: en
• タイトル: Identifying Topic Evolutionary Patterns Based on Component Decomposition of Temporal Sequence
• 言語: eng
• 資源タイプ / 資源タイプ識別子: http://purl.org/coar/resource_type/c_18gh
• 資源タイプ: technical report
• 著者所属: Graduate School of Information, Production and Systems ,Waseda University
• 著者所属: Graduate School of Information, Production and Systems ,Waseda University
• 著者所属: Graduate School of Information, Production and Systems ,Waseda University
• 著者所属: Graduate School of Information, Production and Systems ,Waseda University
• 著者所属(英): en
• 著者所属(英): Graduate School of Information, Production and Systems ,Waseda University
• 著者所属(英): en
• 著者所属(英): Graduate School of Information, Production and Systems ,Waseda University
• 著者所属(英): en
• 著者所属(英): Graduate School of Information, Production and Systems ,Waseda University
• 著者所属(英): en
• 著者所属(英): Graduate School of Information, Production and Systems ,Waseda University
• 著者名: Yingyi, Zhou; Mizuho, Iwaihara
• 著者名(英): Yingyi, Zhou; Mizuho, Iwaihara
• 論文抄録 / 内容記述タイプ: Other
• 論文抄録 / 内容記述: Wikipedia edit history stores a large number of revisions of each article. Once a big event like a gun shooting happens, articles named after events will be created and edited. As the event evolves, there will be edit bursts and key phrases that are covered in these edit bursts will change over time, which we regard as topic evolutionary patterns. To reveal the evolution of topics, we propose a temporal clustering method based on bursts of phrases in articles. In this paper, we try to simplify the time series by burst detection and component decomposition of temporal sequence. Then in clustering, component pattern vectors of phrases which show high cosine similarity will be clustered and key phrases that can explain burst component patterns in a cluster are selected to indicate the topic of the cluster. We solve the problem by temporal information, instead of traditional semantic information. In experiments, we selected the revisions of two article categories. The results of experiments show that the proposed method can identify evolutionary topic temporal patterns reasonably and effectively.
• 論文抄録(英) / 内容記述タイプ: Other
• 論文抄録(英) / 内容記述: Wikipedia edit history stores a large number of revisions of each article. Once a big event like a gun shooting happens, articles named after events will be created and edited. As the event evolves, there will be edit bursts and key phrases that are covered in these edit bursts will change over time, which we regard as topic evolutionary patterns. To reveal the evolution of topics, we propose a temporal clustering method based on bursts of phrases in articles. In this paper, we try to simplify the time series by burst detection and component decomposition of temporal sequence. Then in clustering, component pattern vectors of phrases which show high cosine similarity will be clustered and key phrases that can explain burst component patterns in a cluster are selected to indicate the topic of the cluster. We solve the problem by temporal information, instead of traditional semantic information. In experiments, we selected the revisions of two article categories. The results of experiments show that the proposed method can identify evolutionary topic temporal patterns reasonably and effectively.
• 書誌レコードID / 収録物識別子タイプ: NCID
• 書誌レコードID / 収録物識別子: AN10112482
• 書誌情報: 研究報告データベースシステム（DBS）; 巻 2019-DBS-170, 号 8, p. 1-6, 発行日 2019-12-16
• ISSN / 収録物識別子タイプ: ISSN
• ISSN / 収録物識別子: 2188-871X
• Notice: SIG Technical Reports are nonrefereed and hence may later appear in any journals, conferences, symposia, etc.
• 出版者 / 言語: ja
• 出版者: 情報処理学会