Learning Distributed Representations of Texts and Entities from Knowledge Base
About
We describe a neural network model that jointly learns distributed representations of texts and knowledge base (KB) entities. Given a text in the KB, we train our proposed model to predict entities that are relevant to the text. Our model is designed to be generic with the ability to address various NLP tasks with ease. We train the model using a large corpus of texts and their entity annotations extracted from Wikipedia. We evaluated the model on three important NLP tasks (i.e., sentence textual similarity, entity linking, and factoid question answering) involving both unsupervised and supervised settings. As a result, we achieved state-of-the-art results on all three of these tasks. Our code and trained models are publicly available for further academic research.
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Sentence Relatedness | STS 2014 | News Spearman0.69 | 30 | |
| Factoid Question Answering | History factoid QA (test) | Accuracy94.7 | 25 | |
| Entity Linking | TAC-KBP 2010 (test) | Accuracy87.7 | 16 | |
| Factoid Question Answering | Quiz Bowl Literature category | Accuracy95.1 | 16 | |
| Named Entity Disambiguation | TAC 2010 | Micro Accuracy87.7 | 12 | |
| Semantic Relatedness | SICK | Pearson r0.73 | 12 | |
| Entity Linking | CoNLL (test) | Micro Accuracy94.7 | 11 | |
| Factoid Question Answering | Literature factoid QA (test) | Accuracy95.1 | 9 | |
| Entity Linking | CoNLL-Aida (test) | Accuracy94.3 | 8 | |
| Entity Disambiguation | CoNLL table P (test) | Accuracy94.3 | 7 |