Word-Entity Duet Representations for Document Ranking

Abstract

This paper presents a word-entity duet framework for utilizing knowledge bases in ad-hoc retrieval. In this work, the query and documents are modeled by word-based representations and entity-based representations. Ranking features are generated by the interactions between the two representations, incorporating information from the word space, the entity space, and the cross-space connections through the knowledge graph. To handle the uncertainties from the automatically constructed entity representations, an attention-based ranking model AttR-Duet is developed. With back-propagation from ranking labels, the model learns simultaneously how to demote noisy entities and how to rank documents with the word-entity duet. Evaluation results on TREC Web Track ad-hoc task demonstrate that all of the four-way interactions in the duet are useful, the attention mechanism successfully steers the model away from noisy entities, and together they significantly outperform both word-based and entity-based learning to rank systems.

References in corpus (4)

• Proceedings of the 29th International Conference on Machine Learning (ICML-12)
• A Deep Relevance Matching Model for Ad-hoc Retrieval
• End-to-End Neural Ad-hoc Ranking with Kernel Pooling
• Fast and accurate annotation of short texts with Wikipedia pages

Cited by in corpus (25)

• XLNet: Generalized Autoregressive Pretraining for Language Understanding
• Ad Hoc Table Retrieval using Semantic Similarity
• Representation Learning for Natural Language Processing
• REL: An Entity Linker Standing on the Shoulders of Giants
• Bringing order into the realm of Transformer-based language models for artificial intelligence and law
• Knowledge-Aware Language Model Pretraining
• Neural ranking models for document retrieval
• Towards Better Text Understanding and Retrieval through Kernel Entity Salience Modeling
• Entity-aware Transformers for Entity Search
• EntQA: Entity Linking as Question Answering
• Multi-Cast Attention Networks for Retrieval-based Question Answering and Response Prediction
• Multi-step Entity-centric Information Retrieval for Multi-Hop Question Answering
• Selective Weak Supervision for Neural Information Retrieval
• Entity-Duet Neural Ranking: Understanding the Role of Knowledge Graph Semantics in Neural Information Retrieval
• Semantic Table Retrieval using Keyword and Table Queries
• A Deep Look into Neural Ranking Models for Information Retrieval
• Conversational Entity Linking: Problem Definition and Datasets
• Personal Entity, Concept, and Named Entity Linking in Conversations
• Entity Set Search of Scientific Literature: An Unsupervised Ranking Approach
• THUIR@COLIEE-2020: Leveraging Semantic Understanding and Exact Matching for Legal Case Retrieval and Entailment
• Overcoming low-utility facets for complex answer retrieval
• QDER: Query-Specific Document and Entity Representations for Multi-Vector Document Re-Ranking
• Few-Shot Text Ranking with Meta Adapted Synthetic Weak Supervision
• Learning to Ask: Question-based Sequential Bayesian Product Search
• Semantic Driven Fielded Entity Retrieval