Automatic Alignment between TV News and Newspaper Articles by Maximum Length String Matching between Captions and Article Texts
-
- TSUNODA Tatsuhiko
- Department of Electronics and Communication, Kyoto University
-
- OOISHI Takumi
- Department of Electronics and Communication, Kyoto University
-
- WATANABE Yasuhiko
- Department of Electronics and Informatics, Ryukoku University
-
- NAGAO Makoto
- Department of Electronics and Communication, Kyoto University
Bibliographic Information
- Other Title
-
- キャプションと記事テキストの最長一致文字列照合による報道番組と新聞記事との対応づけの自動化
Search this article
Description
We propose a method of automatic alignment of newspaper articles with corresponding TV news. The method extracts maximum length strings matched between the articles and the ciption texts. Then it calculates similarity and picks up the nearest article if the similarity exceeds a given threshold. The similarity is based on the information of the string, i.e. its length, position and frequency in the text. By adjusting the weighting values, our method achieved l00% recall and 93.2% precision for learning samples, and 98.0% recall and 77.8% (84.5% if the threshold was readjusted) precision for unseen samples. We also clarified the problem of excessive weighting for long strings,
Journal
-
- IPSJ SIG Notes
-
IPSJ SIG Notes 43 17-24, 1996-09-12
Information Processing Society of Japan (IPSJ)
- Tweet
Details 詳細情報について
-
- CRID
- 1573387452012790272
-
- NII Article ID
- 110002934137
-
- NII Book ID
- AN10114171
-
- Text Lang
- ja
-
- Data Source
-
- CiNii Articles
