Towards Understanding and Answering Multi-Sentence Recommendation Questions on Tourism

TL;DR

This paper presents a system for understanding and answering complex multi-sentence recommendation questions in tourism using a novel semantic query language and semi-supervised learning models, achieving significant performance improvements.

Contribution

It introduces a new pipeline with a semantic query language and semi-supervised models for complex recommendation questions in tourism, outperforming baselines.

Findings

Best model is semi-supervised BiDiLSTM+CRF with features and CCM constraints.

System achieves up to 20 points higher accuracy.

System achieves up to 17 points higher recall.

Abstract

We introduce the first system towards the novel task of answering complex multisentence recommendation questions in the tourism domain. Our solution uses a pipeline of two modules: question understanding and answering. For question understanding, we define an SQL-like query language that captures the semantic intent of a question; it supports operators like subset, negation, preference and similarity, which are often found in recommendation questions. We train and compare traditional CRFs as well as bidirectional LSTM-based models for converting a question to its semantic representation. We extend these models to a semisupervised setting with partially labeled sequences gathered through crowdsourcing. We find that our best model performs semi-supervised training of BiDiLSTM+CRF with hand-designed features and CCM(Chang et al., 2007) constraints. Finally, in an end to end QA system, our answering component converts our question representation into queries fired on underlying knowledge sources. Our experiments on two different answer corpora demonstrate that our system can significantly outperform baselines with up to 20 pt higher accuracy and 17 pt higher recall.