Conversational Question Answering over Knowledge Graphs with Transformer and Graph Attention Networks

TL;DR

This paper introduces LASAGNE, a novel neural model combining transformers and graph attention networks for complex conversational question answering over knowledge graphs, outperforming existing methods on standard benchmarks.

Contribution

LASAGNE is the first model to integrate transformer architecture with Graph Attention Networks for multi-task semantic parsing in conversational QA.

Findings

LASAGNE outperforms baselines on all question types.

F1-score increases by over 20% on some question types.

Effective entity recognition and linking in conversational context.

Abstract

This paper addresses the task of (complex) conversational question answering over a knowledge graph. For this task, we propose LASAGNE (muLti-task semAntic parSing with trAnsformer and Graph atteNtion nEtworks). It is the first approach, which employs a transformer architecture extended with Graph Attention Networks for multi-task neural semantic parsing. LASAGNE uses a transformer model for generating the base logical forms, while the Graph Attention model is used to exploit correlations between (entity) types and predicates to produce node representations. LASAGNE also includes a novel entity recognition module which detects, links, and ranks all relevant entities in the question context. We evaluate LASAGNE on a standard dataset for complex sequential question answering, on which it outperforms existing baseline averages on all question types. Specifically, we show that LASAGNE improves the F1-score on eight out of ten question types; in some cases, the increase in F1-score is more than 20% compared to the state of the art.