Abstract:Web search provides a promising way for people to obtain information and has been extensively studied. With the surgence of deep learning and large-scale pre-training techniques, various neural information retrieval models are proposed and they have demonstrated the power for improving search (especially, the ranking) quality. All these existing search methods follow a common paradigm, i.e. index-retrieve-rerank, where they first build an index of all documents based on document terms (i.e., sparse inverted index) or representation vectors (i.e., dense vector index), then retrieve and rerank retrieved documents based on similarity between the query and documents via ranking models. In this paper, we explore a new paradigm of information retrieval with neither sparse nor dense index but only a model. Specifically, we propose a pre-training model-based IR system called DynamicRetriever. As for this system, the training stage embeds the token-level and document-level information (especially, document identifiers) of the corpus into the model parameters, then the inference stage directly generates document identifiers for a given query. Compared with existing search methods, the model-based IR system has two advantages: i) it parameterizes the traditional static index with a pre-training model, which converts the document semantic mapping into a dynamic and updatable process; ii) with separate document identifiers, it captures both the term-level and document-level information for each document. Extensive experiments conducted on the public search benchmark MS MARCO verify the effectiveness and potential of our proposed new paradigm for information retrieval.

Microblog Retrieval Based on Concept-Enhanced Pre-Training Model

A Content-Based Intelligent Ranking Model for Microblog

Learning to Rank Microblog Posts for Real-Time Ad-Hoc Search.

Feature Analysis in Microblog Retrieval Based on Learning to Rank

Leveraging Semantic and Lexical Matching to Improve the Recall of Document Retrieval Systems: A Hybrid Approach

A Hyperlink-Extended Language Model for Microblog Retrieval

Simple Attention-Based Representation Learning for Ranking Short Social Media Posts

Multi-Perspective Relevance Matching with Hierarchical ConvNets for Social Media Search

Language modeling for Microblog retrieval: combine multiple-Bernoulli model and temporal prior for tweets rank

Ranking model selection and fusion for effective microblog search

Microblog Dimensionality Reduction - A Deep Learning Approach.

DynamicRetriever: A Pre-training Model-based IR System with Neither Sparse nor Dense Index

Feedback Model for Microblog Retrieval.

Compact Indexing and Judicious Searching for Billion-Scale Microblog Retrieval.

Online Multi-Core Ranking Model for Image Retrieval

Deep Learning in Exploring Semantic Relatedness for Microblog Dimensionality Reduction

Hybrid Pseudo-Relevance Feedback for Microblog Retrieval

ImageSaker : A Semantic-based Image Retrieval System Refining with Concept Model

User Embedding for Scholarly Microblog Recommendation

Microblog sentiment analysis method using BTCBMA model in Spark big data environment

Pre-trained Language Model for Web-scale Retrieval in Baidu Search