Personalizing Search Results Using Hierarchical RNN with Query-aware Attention

  title={Personalizing Search Results Using Hierarchical RNN with Query-aware Attention},
  author={Songwei Ge and Zhicheng Dou and Zhengbao Jiang and Jian-Yun Nie and Ji-Rong Wen},
  journal={Proceedings of the 27th ACM International Conference on Information and Knowledge Management},
  • Songwei Ge, Zhicheng Dou, Ji-Rong Wen
  • Published 17 October 2018
  • Computer Science
  • Proceedings of the 27th ACM International Conference on Information and Knowledge Management
Search results personalization has become an effective way to improve the quality of search engines. [] Key Method To implement these intuitions, in this paper we employ a hierarchical recurrent neural network to exploit such sequential information and automatically generate user profile from historical data. We propose a query-aware attention model to generate a dynamic user profile based on the input query. Significant improvement is observed in the experiment with data from a commercial search engine when…

Figures and Tables from this paper

PSTIE: Time Information Enhanced Personalized Search

PSTIE, a fine-grained Time Information Enhanced model to construct more accurate user interest representations for Personalized Search, is proposed and experiments show that PSTIE can effectively improve the ranking quality over state-of-the-art models.

PSSL: Self-supervised Learning for Personalized Search with Contrastive Sampling

A Personalized Search framework with Self-supervised Learning (PSSL) to enhance data representations and experimental results demonstrate that the proposed model PSSL achieves state-of-the-art performance compared with existing baselines.

Clarifying Ambiguous Keywords with Personal Word Embeddings for Personalized Search

This article proposes a personalized search model with personal word embeddings for each individual user that mainly contain the word meanings that the user already knows and can reflect the user interests and shows results that significantly outperforms state-of-the-art personalization models.

Contrastive Learning of User Behavior Sequence for Context-Aware Document Ranking

This work proposes a method based on contrastive learning, which takes into account the possible variations in user's behavior sequences, and proposes three data augmentation strategies to generate similar variants of user behavior sequences and contrast them with other sequences.

Encoding History with Context-aware Representation Learning for Personalized Search

A query disambiguation model to parse the meaning of current query in multiple stages, endowed with the benefit from transformer on aggregating contextual information, and a personalized language model to predict user intent from existing queries is devised.

Integrating Representation and Interaction for Context-aware Document Ranking

This work proposes a unified context-aware document ranking model which takes full advantage of both representation and interaction, and encodes the session history into a latent representation and uses this representation to enhance the current query and the candidate document.

Enhancing Re-finding Behavior with External Memories for Personalized Search

This paper proposes to construct memory networks (MN) to support the identification of more complex re-finding behavior, and devise two external memories to make an expansion of re- finding based on the query and the document respectively.

IHGNN: Interactive Hypergraph Neural Network for Personalized Product Search

This work proposes a new model IHGNN for personalized product search that resorts to a hypergraph constructed from the historical user-product-query interactions, which could completely preserve ternary relations and express collaborative signal based on the topological structure.

Employing Personal Word Embeddings for Personalized Search

A personalized search model based on personal word embeddings for each user in which the representation of each word is mainly decided by the user's personal data is designed, and can significantly improve state-of-the-art personalization models.

Improving Personalized Search with Dual-Feedback Network

This paper proposes a dual-feedback modeling framework, which integrates multi-granular user feedback information to model the user's current search intention, and proposes a feedback extraction network to refine the dual-feedingback representation in multiple stages.



A Hierarchical Recurrent Encoder-Decoder for Generative Context-Aware Query Suggestion

This work presents a novel hierarchical recurrent encoder-decoder architecture that makes possible to account for sequences of previous queries of arbitrary lengths and is sensitive to the order of queries in the context while avoiding data sparsity.

Deep Learning Powered In-Session Contextual Ranking using Clickthrough Data

This work demonstrates how to generate the semantic features from in-session contextual information with deep learning models, and incorporate these semantic features into the current ranking model to re-rank the results.

Enhancing personalized search by mining and modeling task behavior

A method whereby other users performing similar tasks to the current user and leverage their on-task behavior to identify Web pages to promote in the current ranking yields promising gains in retrieval performance, and has direct implications for improving personalization in search systems.

Personalized document re-ranking based on Bayesian probabilistic matrix factorization

This work considers the searcher's search and page view history, and uses dwell time on clicked URLs when estimating the relevance of documents for a query, and performs Bayesian Probabilistic Matrix Factorization as smoothing to predict the relevance.

To personalize or not to personalize: modeling queries with variation in user intent

Variation in user intent is examined using both explicit relevance judgments and large-scale log analysis of user behavior patterns to identify queries that can benefit from personalization.

Probabilistic models for personalizing web search

This work proposes a generative model of relevance which can be used to infer the relevance of a document to a specific user for a search query, and shows how to learn these profiles from a user's long-term search history.

Temporal Latent Topic User Profiles for Search Personalisation

The experimental results demonstrate that the temporal profiles built using Latent Dirichlet Allocation can significantly improve the ranking quality and show a promising effect of temporal features in correlation with click entropy and query position in a search session.

A large-scale evaluation and analysis of personalized search strategies

It is revealed that personalized search has significant improvement over common web search on some queries but it also has little effect on other queries, and even harms search accuracy under some situations.

Personalized ranking model adaptation for web search

This paper proposes a general ranking model adaptation framework for personalized search that quickly learns to apply a series of linear transformations over the parameters of the given global ranking model such that the adapted model can better fit each individual user's search preferences.

Improving search personalisation with dynamic group formation

This paper proposes a personalisation framework in which a user profile is enriched using information from other users dynamically grouped with respect to an input query, and demonstrates that the framework improves the performance of the web search engine and also achieves better performance than the static grouping method.