Leveraging Semantic Web Search and Browse Sessions for Multi-Turn Spoken Dialog Systems

TL;DR

This paper introduces a novel approach that leverages large-scale web search and browse sessions to improve the training of spoken dialog systems, addressing data scarcity issues and enhancing entity and relation extraction performance.

Contribution

It presents a method to automatically mine behavioral patterns from web logs and translate them into dialog models, improving over existing techniques.

Findings

Session-based models outperform state-of-the-art entity extraction methods.

Improved performance in entity and relation extraction on web search queries.

Demonstrates generalization of web session behaviors to spoken dialog tasks.

Abstract

Training statistical dialog models in spoken dialog systems (SDS) requires large amounts of annotated data. The lack of scalable methods for data mining and annotation poses a significant hurdle for state-of-the-art statistical dialog managers. This paper presents an approach that directly leverage billions of web search and browse sessions to overcome this hurdle. The key insight is that task completion through web search and browse sessions is (a) predictable and (b) generalizes to spoken dialog task completion. The new method automatically mines behavioral search and browse patterns from web logs and translates them into spoken dialog models. We experiment with naturally occurring spoken dialogs and large scale web logs. Our session-based models outperform the state-of-the-art method for entity extraction task in SDS. We also achieve better performance for both entity and relation extraction on web search queries when compared with nontrivial baselines.