Semantic matching is critical in many text applications, including paraphrase identification, information retrieval, question answering, and machine translation. A variety of machine learning techniques have been developed for various semantic matching tasks, referred to as “learning to match”. Recently, deep learning approaches have shown their effectiveness in this area, and a number of methods have been proposed from different aspects of matching. In this tutorial, we will give a systematic and detailed survey on newly developed deep learning technologies for semantic matching. We will focus on the descriptions on the fundamental problems, as well as the novel solutions from bridging the word level semantic gap and conducting sentence level end-to-end semantic matching. We will also discuss the potential applications and future directions of semantic matching for text.


  • Overview of the Semantic Matching for Text
    • Problem Description
    • Learning to Match
    • Deep Learning to Match
  • Word-Level Representation
    • Word Embedding based on Matrix Factorization
    • Neural Word Embedding
    • Semantic Matching using Word Embeddings
  • End-to-End Deep Matching Methods
    • Deep Matching Methods with Single Sentence Representation
    • Deep Matching Methods with Multiple Granularity Sentence Representations
    • Deep Matching Methods with Structure Representation
  • Discussions and Summary


  • Jiafeng Guo

    Jiafeng Guo is an Associate Professor at Institute of Computing Technology, Chinese Academy of Sciences. He has worked on a number of topics related to Web search and data mining, including query representation and understanding, learning to rank, and topic modeling. His current research is focused on representation learning and deep models for information retrieval and information filtering. He has published more than 80 papers in international journals and conferences, including IEEE TKDE, Information Retrieval Journal, Journal of Statistical Mechanics, SIGIR, WWW, CIKM, ICDM, WSDM, and IJCAI. He has won the Best Paper Award in ACM CIKM (2011) and Best Student Paper Award in ACM SIGIR (2012).

  • Yanyan Lan

    Yanyan Lan received the B.E. degree in Statistics from Shandong University, Jinan, China, in 2005 and the Ph.D. degree in Probability and Statistics from the Institute of Applied Mathematics, Academy of Mathematics and System Sciences, Chinese Academy of Sciences, Beijing, China, in 2011. She is currently an Associate Professor in Institute of Computing Technology, Chinese Academy of Sciences. She leads a research group working on Big Data and Machine Learning. Her current research interests include Machine Learning, Web Search and Data Mining, and Big Data Analysis. She has published over 30 papers on top conferences including ICML, NIPS, SIGIR, WWW et al., and the paper entitled “Top-k Learning to Rank: Labeling, Ranking, and Evaluation” has won the Best Student Paper Award of SIGIR 2012.


Slides can be downloaded from here.