ゼロショット学習による文章からのエンティティの性質の発見
Identifying Entity Properties from Text with Zero-shot Learning
概要
Identifying entity properties from text is a task for classifying properties of a focused entity in sentences. We propose a neural network-based model that utilizes a knowledge graph embedding and an entity type information for solving a zero-shot learning problem where a training set is incomplete.
産業界への展開例・適用分野
Identified properties can be utilized in existing search functionalities such as entity cards, web snippets, and document ranking. As a result, this could enable a personalized entity-oriented search.
研究者
| 氏名 | 専攻 | 研究室 | 役職/学年 |
|---|---|---|---|
| Imrattanatrai Wiradee | 社会情報学専攻 | 分散情報システム分野 | 博士3回生 |