Abstract

AbstractWe present the novel task of understanding multi-sentenceentity-seekingquestions (MSEQs), that is, the questions that may be expressed in multiple sentences, and that expect one or more entities as an answer. We formulate the problem of understanding MSEQs as a semantic labeling task over an open representation that makes minimal assumptions about schema or ontology-specific semantic vocabulary. At the core of our model, we use a BiLSTM (bidirectional LSTM) conditional random field (CRF), and to overcome the challenges of operating with low training data, we supplement it by using BERT embeddings, hand-designed features, as well as hard and soft constraints spanning multiple sentences. We find that this results in a 12–15 points gain over a vanilla BiLSTM CRF. We demonstrate the strengths of our work using the novel task of answering real-world entity-seeking questions from the tourism domain. The use of our labels helps answer 36% more questions with 35% more (relative) accuracy as compared to baselines. We also demonstrate how our framework can rapidly enable the parsing of MSEQs in an entirely new domain with small amounts of training data and little change in the semantic representation.

Highlights

  • We are motivated by the goal of building an information agent for tourists – one that would perform various roles of a travel agent, such as helping decide the city to visit, recommending points of interest, finding travel routes, and even creating optimized itineraries

  • We present the first system for the novel task of answering entity-seeking MSRQs from background corpora in the tourism domain

  • Before describing implementation details of our question understanding component, we present some background on Constrained Conditional Models (CCMs)(Chang et al, 2007) and BiDiLSTM+CRF(Huang et al, 2015) as these are at the core of our question understanding component

Read more

Summary

Introduction

We are motivated by the goal of building an information agent for tourists – one that would perform various roles of a travel agent, such as helping decide the city to visit, recommending points of interest, finding travel routes, and even creating optimized itineraries. We focus our paper on questions that are entity-seeking, i.e., expect one or more entities as answer These include the large fraction of tourist questions that ask for hotels, restaurants, points of interest and other services that would serve a user’s specific needs the best. A preliminary analysis of such questions from popular tourism forums reveals that almost all of them contain multiple sentences – they often elaborate on a user’s specific situation before asking their question. We name these MSRQs – multi-sentence recommendation questions. The querying module needs to incorporate the various constructs found in recommendation questions

Contributions
Related Work
System Architecture
RQL Representation
Question Understanding
Background on Sequence Labeling
Semantic Labeling of Questions
Supervised Labeling Conditional Random Field
Crowd-sourced Data Collection
Semi-Supervised Labeling
Evaluation
Results
Answer Generation
Conclusion and Future Work
Full Text
Paper version not known

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call

Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.