| 研究課題/領域番号 |
17H01828
|
|---|
| 研究機関 | 京都大学 |
|---|
研究代表者 |
Adam Jatowt 京都大学, 情報学研究科, 特定准教授 (00415861)
|
|---|
| 研究分担者 |
Zhang Yating 国立研究開発法人理化学研究所, 革新知能統合研究センター, 特別研究員 (30793559)
澄川 靖信 東京理科大学, 理工学部情報科学科, 助教 (70756303)
|
|---|
| 研究期間 (年度) |
2017-04-01 – 2021-03-31
|
| キーワード | computational history / collective memory / text mining / document archives |
|---|
| 研究実績の概要 |
The first year research resulted in analysis of history-related references in Twitter. By analyzing over 1 million tweets that relate to history we have determined the horizon and the focus of collective memories in social media. In another work we have built a system for classifying short descriptions of historical events into 9 distinct event classes using a range of features baed on Wikipedia, VerbNet and word embeddings. This work has appeared in ECIR2017 conference. We have also proposed method for summarizing multiple histories of related entities such as cities in the same country using document summarization and timeline processing methods. This work has been published at HT2017 conference.
|
| 現在までの達成度 (区分) |
現在までの達成度 (区分)
2: おおむね順調に進展している
理由
The project has been proceeding according to the plans and resulted in several publications at prestigious conferences as well as a well-known journal in a digital library field.
|
| 今後の研究の推進方策 |
Our next plans involve finding analogous historical events to present events for improving history education as well as for supporting future forecasting. We will also work on summarizing texts in news archives and on retrieving knowledge from long-term news archives through contrastive and evolutionary summarization. More work will be also done towards studies of collective memories in social media.
|
