{"id":171429,"date":"2015-01-30T16:49:10","date_gmt":"2015-01-30T16:49:10","guid":{"rendered":"https:\/\/www.microsoft.com\/en-us\/research\/project\/dssm\/"},"modified":"2019-08-19T10:45:32","modified_gmt":"2019-08-19T17:45:32","slug":"dssm","status":"publish","type":"msr-project","link":"https:\/\/www.microsoft.com\/en-us\/research\/project\/dssm\/","title":{"rendered":"DSSM"},"content":{"rendered":"<div class=\"asset-content\">The goal of this project is to develop a class of deep\u00a0representation learning models. DSSM stands for Deep Structured Semantic Model, or more general, Deep Semantic Similarity Model. DSSM, developed by the MSR Deep Learning Technology Center(<a class=\"invalidLink\" href=\"https:\/\/www.microsoft.com\/en-us\/research\/group\/deep-learning-group\/\" target=\"_self\" rel=\"noopener noreferrer\">DLTC<\/a>), is a deep neural network (DNN) modeling technique for representing text strings (sentences, queries, predicates, entity mentions, etc.) in a\u00a0continuous semantic space and\u00a0modeling semantic similarity between two text strings (e.g., Sent2Vec). DSSM has wide applications including information retrieval and\u00a0web search ranking (<a class=\"invalidLink\" href=\"https:\/\/www.microsoft.com\/en-us\/research\/publication\/learning-deep-structured-semantic-models-for-web-search-using-clickthrough-data\/\" target=\"_self\" rel=\"noopener noreferrer\">Huang et al. 2013<\/a>; <a class=\"invalidLink\" href=\"https:\/\/www.microsoft.com\/en-us\/research\/publication\/learning-semantic-representations-using-convolutional-neural-networks-for-web-search\/\" target=\"_self\" rel=\"noopener noreferrer\">Shen et al. 2014a<\/a>,<a class=\"invalidLink\" href=\"https:\/\/www.microsoft.com\/en-us\/research\/publication\/a-latent-semantic-model-with-convolutional-pooling-structure-for-information-retrieval\/\" target=\"_self\" rel=\"noopener noreferrer\">2014b <\/a>; <a href=\"https:\/\/www.microsoft.com\/en-us\/research\/publication\/deep-sentence-embedding-using-long-short-term-memory-networks-analysis-application-information-retrieval\/\">Palangi et al.2016<\/a>), ad selection\/relevance, contextual entity search and interestingness tasks (<a class=\"invalidLink\" href=\"\" target=\"_self\" rel=\"noopener noreferrer\">Gao et al. 2014a<\/a>), question answering (<a class=\"invalidLink\" href=\"https:\/\/www.microsoft.com\/en-us\/research\/publication\/semantic-parsing-for-single-relation-question-answering\/\" target=\"_self\" rel=\"noopener noreferrer\">Yih et al., 2014<\/a>), knowledge inference (<a class=\"msr-external-link glyph-append glyph-append-open-in-new-tab glyph-append-xsmall\" href=\"http:\/\/arxiv-web3.library.cornell.edu\/abs\/1412.6575\" target=\"_blank\" rel=\"noopener noreferrer\">Yang et al., 2014<span class=\"sr-only\"> (opens in new tab)<\/span><\/a>), image captioning (<a class=\"msr-external-link glyph-append glyph-append-open-in-new-tab glyph-append-xsmall\" href=\"http:\/\/arxiv.org\/abs\/1411.4952\" target=\"_blank\" rel=\"noopener noreferrer\">Fang et al., 2014<span class=\"sr-only\"> (opens in new tab)<\/span><\/a>),\u00a0and machine translation (<a class=\"invalidLink\" href=\"https:\/\/www.microsoft.com\/en-us\/research\/publication\/learning-continuous-phrase-representations-for-translation-modeling\/\" target=\"_self\" rel=\"noopener noreferrer\">Gao et al., 2014b<\/a>) etc. DSSM can be used to develop latent semantic models that project entities of different types (e.g., queries and documents) into a common low-dimensional semantic space for a variety of machine learning tasks such as ranking and classification. For example, in web search ranking, the relevance of a document given a query can be readily computed as the distance between them in that space.\u00a0With the latest GPUs from Nvidia, we are able to train our models on billions of\u00a0words.\u00a0Readers that are interested in deep learning for text processing may refer to our recent tutorial (<a class=\"invalidLink\" href=\"https:\/\/www.microsoft.com\/en-us\/research\/publication\/deep-learning-for-natural-language-processing-theory-and-practice-tutorial\/\" target=\"_self\" rel=\"noopener noreferrer\">He et al., 2014<\/a>), (<a href=\"https:\/\/www.microsoft.com\/en-us\/research\/publication\/introduction-deep-learning-natural-language-processing-tutorial-deeplearning2017-summer-school-bilbao-2\/#\">Gao 2017<\/a>).<\/div>\n<div id=\"en-usprojectsdssmdefault\" class=\"page-content\">\n<p>We released the <a href=\"https:\/\/www.microsoft.com\/en-us\/download\/details.aspx?id=52365\">predictors and trained model files <\/a>of the DSSM (also a.k.a. Sent2Vec).<\/p>\n<\/div>\n","protected":false},"excerpt":{"rendered":"<p>The goal of this project is to develop a class of deep\u00a0representation learning models. DSSM stands for Deep Structured Semantic Model, or more general, Deep Semantic Similarity Model. DSSM, developed by the MSR Deep Learning Technology Center(DLTC), is a deep neural network (DNN) modeling technique for representing text strings (sentences, queries, predicates, entity mentions, etc.) [&hellip;]<\/p>\n","protected":false},"featured_media":0,"template":"","meta":{"msr-url-field":"","msr-podcast-episode":"","msrModifiedDate":"","msrModifiedDateEnabled":false,"ep_exclude_from_search":false,"_classifai_error":"","footnotes":""},"research-area":[13556,13545,13555],"msr-locale":[268875],"msr-impact-theme":[],"msr-pillar":[],"class_list":["post-171429","msr-project","type-msr-project","status-publish","hentry","msr-research-area-artificial-intelligence","msr-research-area-human-language-technologies","msr-research-area-search-information-retrieval","msr-locale-en_us","msr-archive-status-active"],"msr_project_start":"1\/30\/2015","related-publications":[167045,168020,418055,165215,166530,166529,166513,168302,365066],"related-downloads":[],"related-videos":[],"related-groups":[],"related-events":[],"related-opportunities":[],"related-posts":[],"related-articles":[],"tab-content":[],"slides":[],"related-researchers":[{"type":"user_nicename","display_name":"Jianfeng Gao","user_id":32246,"people_section":"Group name 1","alias":"jfgao"}],"msr_research_lab":[],"msr_impact_theme":[],"_links":{"self":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-project\/171429","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-project"}],"about":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/types\/msr-project"}],"version-history":[{"count":8,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-project\/171429\/revisions"}],"predecessor-version":[{"id":604248,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-project\/171429\/revisions\/604248"}],"wp:attachment":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/media?parent=171429"}],"wp:term":[{"taxonomy":"msr-research-area","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/research-area?post=171429"},{"taxonomy":"msr-locale","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-locale?post=171429"},{"taxonomy":"msr-impact-theme","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-impact-theme?post=171429"},{"taxonomy":"msr-pillar","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-pillar?post=171429"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}