論 文Papers

CONFERENCE (DOMESTIC)

ナレッジベースを用いた記事の主題地域推定システム

佐々木 明, 大倉 俊平

2020年度 人工知能学会全国大会 (第34回) (JSAI 2020), 2020/6

Category:

自然言語処理 (Natural Language Processing) データサイエンス (Data Science) セマンティック・ウェブ (Semantic Web)

Abstract:
Predicting relevant locations from news articles is a task that finds applications in many web-related services. In this paper, we propose a simple dictionary-based approach to predict relevant locations from news article text. It is a real-world application of knowledge base-related technologies. By leveraging a subgraph of knowledge base that contains information about locations, our system achieves high precision and recall compared to baseline methods, that makes predictions based on direct mentions of administrative areas. The method is evaluated using a dataset that consists of news articles and salient prefectures for each article hand-labeled by professional annotators. The system described in this paper is actually used in one of the services provided by Yahoo! Japan.
Download:

ナレッジベースを用いた記事の主題地域推定システム(外部サイト/External Site Link)