Joint Intent Detection and Slot Filling with Wheel-Graph Attention Networks

TL;DR

This paper introduces a novel wheel-graph attention network for joint intent detection and slot filling, effectively modeling interrelated connections and leveraging BERT to enhance spoken language understanding performance.

Contribution

The paper proposes a new Wheel-GAT model that explicitly models interrelations between intent and slot nodes for improved SLU tasks, integrating BERT for further performance gains.

Findings

Outperforms multiple baselines on public datasets

Model benefits from BERT integration

Effective modeling of intent-slot interrelations

Abstract

Intent detection and slot filling are two fundamental tasks for building a spoken language understanding (SLU) system. Multiple deep learning-based joint models have demonstrated excellent results on the two tasks. In this paper, we propose a new joint model with a wheel-graph attention network (Wheel-GAT) which is able to model interrelated connections directly for intent detection and slot filling. To construct a graph structure for utterances, we create intent nodes, slot nodes, and directed edges. Intent nodes can provide utterance-level semantic information for slot filling, while slot nodes can also provide local keyword information for intent. Experiments show that our model outperforms multiple baselines on two public datasets. Besides, we also demonstrate that using Bidirectional Encoder Representation from Transformer (BERT) model further boosts the performance in the SLU task.