NE-Table: A Neural key-value table for Named Entities

TL;DR

This paper introduces NE-Table, a neural key-value table that improves handling of Named Entities, especially OOV and rare ones, across various NLP tasks like question-answering and dialog systems.

Contribution

The paper presents NE-Table, a novel neural key-value table structure that enhances the processing of Named Entities, including out-of-vocabulary ones, in neural NLP models.

Findings

Effective in handling OOV and rare NEs in multiple NLP tasks

Improves performance on structured QA, goal-oriented dialogs, and reading comprehension

Demonstrates robustness with extended and OOV datasets

Abstract

Many Natural Language Processing (NLP) tasks depend on using Named Entities (NEs) that are contained in texts and in external knowledge sources. While this is easy for humans, the present neural methods that rely on learned word embeddings may not perform well for these NLP tasks, especially in the presence of Out-Of-Vocabulary (OOV) or rare NEs. In this paper, we propose a solution for this problem, and present empirical evaluations on: a) a structured Question-Answering task, b) three related Goal-Oriented dialog tasks, and c) a Reading-Comprehension task, which show that the proposed method can be effective in dealing with both in-vocabulary and OOV NEs. We create extended versions of dialog bAbI tasks 1,2 and 4 and OOV versions of the CBT test set available at - https://github.com/IBM/ne-table-datasets.