Speaker-Sensitive Dual Memory Networks for Multi-Turn Slot Tagging

TL;DR

This paper introduces a speaker-sensitive dual memory network architecture that improves multi-turn slot tagging by effectively encoding dialog history based on speaker roles, leading to better contextual understanding.

Contribution

It proposes a novel neural model that encodes utterances differently depending on the speaker, enhancing multi-turn slot tagging accuracy.

Findings

Significant performance improvement over state-of-the-art models

Effective encoding of speaker-specific contextual information

Validated on real Microsoft Cortana data

Abstract

In multi-turn dialogs, natural language understanding models can introduce obvious errors by being blind to contextual information. To incorporate dialog history, we present a neural architecture with Speaker-Sensitive Dual Memory Networks which encode utterances differently depending on the speaker. This addresses the different extents of information available to the system - the system knows only the surface form of user utterances while it has the exact semantics of system output. We performed experiments on real user data from Microsoft Cortana, a commercial personal assistant. The result showed a significant performance improvement over the state-of-the-art slot tagging models using contextual information.