Description: With the increasing prevalence of intelligent interaction platforms or products such as personal assistants, smart speakers and voice-controlled assistants in cars, the new generation of AI-enabled intelligent interaction systems attracts more and more attentions from both academy and industry. As a crucial component of such systems, natural language processing technology has also made great progress in recent years. This workshop will focus on the basic methods, core technologies and practical experiences of natural language processing technologies, particularly on intelligent interactive platforms and scenarios.
We welcome the experts both from academia and from industry to participate in the discussion and the idea exchange on“NLP for intelligent interaction”.

Talk 1: Tencent Machine Translation Systems: Interaction and Application, by Mingxuan Wang

Abstract: Machine Translation is an important part of the AI-strategy of Tencent which aims at solving barriers of communication and helping people to access information from all over the world freely. Every day, Tencent Machine Translation platforms supported more than 400 million translation requests via various interactive approaches including text, voice, and image. In this talk, we will first present the basic architecture of Tencent Machine Translation Systems, and then introduce some practical experience of building our real time online system, particularly on intelligent interactive technologies.

About the Speaker: Mingxuan Wang is currently a senior researcher at Mobile Internet Group, Tencent Technology. Before joining Tencent in July 2017, he visited Noah’s ark Lab from 2014 to 2015 and IDL lab during 2016 as an assistant researcher. Wang received his PHD from Institute of Computing Technology, Chinese Academy of Sciences. He has published more than 10 academic papers on top conference including ACL, EMNLP, AAAI, IJCAI and COLING. His research interests lie in Natural Language Processing, with a focus on Neural Machine Translation.

Talk 2: Knowledge Engine is Everywhere in Tencent QQ Browser, by Yancheng He

Abstract: Knowledge is an important infrastructure for many artificial intelligence tasks. Tencent knowledge engine is a system covering knowledge gathering, mining, reasoning and its applications. Currently it has already played a vital part in many applications in Tencent, such as search engine, recommendation systems, dialogue systems and event mining. This talk starts from the introduction of the Tencent knowledge engine, and then presents its applications in Tencent.

About the Speaker: Yancheng He is a principal researcher at Mobile Internet Group, Tencent Inc. He is currently the director of the big data research center, Mobile Browser Product Department. His research interest covers a wide range of artificial intelligence domain, including knowledge engine, user understanding, and natural language processing. He received his master's degree from Harbin Institute of Technology. He has more than 10 years of experience in search engine, computing advertising and big data.

Talk 3: End-to-end Task-oriented Dialog Systems, by Wanxiang Che

Abstract: Task-oriented dialog systems have been successfully applied in many applications such as virtual personal assistants. These systems allow interaction between users via natural language to help users to complete specific tasks. Traditional task-oriented dialog systems are pipelines that consist of several components. In this talk, we will briefly discuss each component and we will focus on building end-to-end task-oriented dialog systems based on these components and make the systems end-to-end trainable. Recently, a trend in dialog system is that the boundary between chat-bots and task-oriented bots are becoming blurred. In this talk, we will also discuss some neural architectures, mainly seq2seq based architectures, that are not based on pipeline modules for task-oriented dialog systems. These systems try to build dialog systems without the use of domain-specific data for components of the pipeline systems and some directly retrieve entities from knowledge base without breaking differentiability, which makes training simpler and which is easy to be applied to different domains. We will also introduce our recent work that brings dialog state tracking, a core component of pipelines, into seq2seg learning and shows how it helps retrieve entities from knowledge base and generation.

About the Speaker: Dr. Wanxiang Che, professor of school of computer science and technology at Harbin Institute of Technology (HIT) and visiting scholar of Stanford University (at NLP group in 2012). His main research area lies in Natural Language Processing (NLP). He currently leads a national natural science foundation of China, a national 973 and a number of research projects. He has published more than 40 papers in high level journals and conferences and published two textbooks. He and his team have achieved good results in a number of international technical evaluations, such as the first place of CoNLL 2009 and the fourth place of CoNLL 2017. He was an area co-chair of ACL 2016, publication co-chairs of ACL 2015 and EMNLP 2011. The Language Technology Platform (LTP), an open source Chinese NLP system he leads to develop, has been authorized to more than 600 institutes and individuals including Baidu, Tencent and so on. He achieved the outstanding paper award honorable mention of AAAI 2013, the first prize of technological progress award in Heilongjiang province in 2016, Google focused research award in 2015 and 2016, the first prize of Hanwang youth innovation award and first prize of the Qian Weichan Chinese information processing science and technology award in 2010.

Talk 4: Joint Models for Goal-Driven Dialogue, by Xiaojie Wang

Abstract: Traditional goal-driven dialogue system has been widely studied in a pipeline way, where different sub-tasks in the system are modeled separately. It suffers from error accumulations and some other problems. Joint modeling two or more sub-tasks can alleviate the problems efficiently. This talk introduces some recent work on joint models for goal-driven dialogue, including joint models of sub-tasks in the modules of natural language understanding and dialogue management, as well as joint models across the modules.

About the Speaker: Xiaojie Wang is professor at School of Computer Science, Beijing University of Posts and Telecommunications. He is director of department of artificial intelligence, director of center for intelligence science and technology. He serves as director of Committee of Natural Language Understanding, Chinese Association of Artificial Intelligence. His research interests include natural language processing, man-machine dialogue, multimodal computing.

Talk 5: Toward Diverse Text Generation with Inverse Reinforcement Learning, by Xipeng Qiu

Abstract: Text generation is a crucial task in NLP. Recently, several adversarial generative models have been proposed to improve the exposure bias problem in text generation. Though these models gain great success, they still suffer from the problems of reward sparsity and mode collapse. In this talk, we employ inverse reinforcement learning (IRL) for text generation to address these two problems. Specifically, the IRL framework learns a reward function on training data, and then an optimal policy to maximize the expected total reward. Similar to the adversarial models, the reward and policy function in IRL are optimized alternately. Our method has two advantages: (1) the reward function can produce more dense reward signals. (2) the generation policy, trained by ‘entropy regularized’ policy gradient, encourages to generate more diversified texts.

About the Speaker: Xipeng Qiu is an associate professor and Ph.D. supervisor at the School of Computer Science, Fudan University. He received B.S. and Ph.D. degrees from Fudan University. His major research areas include artificial intelligence, machine learning, deep learning, and natural language processing. He has published more than 50 papers in the top journals and conferences (ACL/EMNLP/IJCAI/AAAI, etc.) He is lead developer of the open source project FudanNLP for Chinese language processing. He was selected for the Young Elite Scientists Sponsorship Program by China Association for Science and Technology (CAST) and won the Outstanding Paper Award of ACL 2017.

Return

Description: A human-computer dialogue system is a computer system intended to coherently converse with a human. The computer should be able to model a human dialogue participant, conversing in a similar linguistic style, discussing the topics presented, and even teaching the human participant using knowledge acquired. With the progress in many relevant perceptual and cognitive AI technologies, more advanced intelligent human-computer dialogue systems have been developed, from typical task-completion conversational systems to various intelligent personal assistants, to most recent social chatbots. Meanwhile, lots of new challenges have also come out. The dialogue system not only needs to solve all the questions the users might have, but also, to dynamically recognize user’s emotion and engage the user throughout long conversations with appropriate interpersonal responses. So, building a dialogue system with “temperature” becomes necessary. What’s more, to communicate effectively, computer also needs to interact with users through multiple modalities, including text, speech, and vision. So multimodal intelligence is another challenge. In this workshop, we will invite researchers to share the latest progress in human-computer dialogue systems.

About the Chair: Dr. Xiaodong He is the Technical Vice President and the Deputy Managing Director of JD AI Research and Head of the Deep learning, NLP and Speech Lab. He is also an Affiliate Professor at the University of Washington (Seattle), serves in doctoral supervisory committees. Before joining JD.com, He was with Microsoft for about 15 years, served as Principal Researcher and Research Manager of the DLTC at Microsoft Research, Redmond.

Talk 1: Title: NLP advances in JD AI, by Bowen Zhou

Abstract: In this talk, I will introduce the recent advances at JD AI that specifically address the business need of JD.COM. First, I will give you a short broad view on our whole AI research, including interesting work on various research fields such as Computer Vision, Speech Recognition, Machine Learning, and Natural Language Processing (NLP). Then, I will focus on two specific aspects of NLP applications: 1) let AI understand human, such as semantic parsing, intent classification, and sentiment analysis; 2) let AI be understood as well, such as natural language generation, topic expansion, summarization and emotional dialog generation. Finally, I will dive into the very challenging problem -- Natural Language Generation (NLG) and share with you some of our latest novel approaches. Besides, I will also show how we apply this cutting-edge technology into JD’s products successfully, such as poetry generation, product description generation and dialogue response generation etc. We will demonstrate on these front-ends NLP has greatly helped our business to better serve our customers in real e-commerce scenario.

About the Speaker: Dr. Bowen Zhou is the Vice President of JD, head of Artificial Intelligence Platform & Research of JD.com . Before joining JD.com, Bowen was with IBM for almost 15 years. In April last year, Bowen was appointed as the Chief Scientist of Watson Group, where he was responsible to lead and align Watson Group’s science agenda with IBM’s technical strategy and IBM Research’s cognitive computing and artificial intelligence agenda. Bowen is a technologist and business leader of human language technologies, machine learning, and artificial intelligence.

Talk 2: The development of human-computer dialogue and its applications, by Xiaoyan Zhu

Abstract: With the popularity of deep learning algorithms, research in artificial intelligence has gradually developed from perceptive intelligence to cognitive intelligence. Human-computer dialogue is one of aspects indicating the cognitive machine intelligence, which explains why it attracts more and more attention and is moving towards applications. This report will discuss the advantages and disadvantages of deep-learning-model-based algorithms in theoretical research, and explore key issues and directions for future research. From application perspective, this report will also discuss, the goal of human-computer dialogue, the methodologies to achieve that goal and other problems in the process of system development.

About the Speaker: Prof. Xiaoyan Zhu is the head of State key Lab of Intelligent Technology and Systems, Tsinghua University. Since 1993, Prof. Zhu has been on the faculty of Department of Computer Science and Technology, Tsinghua University. She received Ph.D. degree at Nagoya Institute of Technology in 1990. Currently her research interests are focus on intelligent information processing, internet information acquisition, and conversational system. She has successfully conducted the research supported by National Basic Research Program (973 program), National High Technology Research and Development Program of China (863 program), and National Natural Science Foundation of China (NSFC). She has got Okawa award, Japan, 2014, Google research award, 2012, 2014, best paper award, COLING 2011, best student paper award, ACL 2012, and best student paper award, SDM 2014, respectively. Prof. Zhu has authored or co-authored more than 100 papers in the top scientific journals such IEEE Tran.on SMC, Journal of Knowledge and Information Systems, Communications of the ACM, and conference proceedings SIG KDD，IJCAI, SIGIR, ACL, AAAI, ICDM, COLING, SDM, and CIKM.

Talk 3: Emotion cognition and recognition of the sounds, by Haifeng Li

Abstract: Emotions are the basis of the social interactions and cognitive activities of human beings. With the help of effective emotional interactions, the user experience of human-computer interaction (HCI) will be greatly enhanced. Thus, more natural HCI remains a top issue of the field of artificial intelligence (AI).
Auditory and vision are two main ways for human to receive emotions. Due to the structure of human hearing system, studies in auditory emotion cognition and recognition are more complicated and started relatively late. Most of them are limited in music research. It is still uncertain in the occurrence mechanism, triggering conditions, influencing factors and cognitive principles of auditory emotions. Therefore, the depth and detailed analysis is in high demand.
Traditional methods of sound emotion recognition are basically based on statistics and machine learning, which limit the development of relevant studies. It is proven that human brain maintains various region and neural pathway for different emotions. Therefore, the most effective methods are from the data processing mechanism of brain emotion system. That is studying the cognition or calculation of auditory emotions with combining with the analyzing and recognizing methods of emotions in sound.
Firstly, the talk is based on the cognitive experimental work of auditory emotions. Then, the current status and problems of cognition and recognition of auditory emotion are presented with other relevant research results. In addition, aimed at several problems, such as the design of auditory cognitive experiments, the relationship between auditory physical features and their emotional response, and auditory emotional cognition in the application of emotional computing, related research work and research results are introduced. Then multi-level auditory cognitive experimental paradigm is proposed based on auditory attribute complexity and semantic complexity. Then, long-term EEG signal analysis and evaluation method based on event-related potential (ERP) and spectrum estimation is introduced. Finally, the prospect for the future study of auditory emotion cognition and recognition are discussed.

About the Speaker: Prof. Haifeng Li is the IEEE member, CCF senior member (China Computer Federation) and professor at the Institute of Intelligent Human-Computer Interaction, School of Computer Science and Technology, Harbin Institute of Technology (HIT). His research fields include: Intelligent Information Processing Technology, Brain Machine Interaction Technology, and Artificial Intelligence and Cognition Science. He undertakes two projects of National Natural Science Foundation, one project of National High-Tech Foundation (863) and several projects of Provincial and Ministry Science Foundation. He published 2 books and over 100 papers in journals and conferences at home and abroad.

Talk 4: Topic-driven Chinese Discourse Machine Reading Comprehension, by Zhou Guodong

Abstract: Currently, the research on NLP focuses on lexical and syntactic levels, and lacks theoretical and methodological systems towards efficient analysis and deep understanding of discourse due to much less research on inherent regulations in discourse. This largely restricts discourse-level research and its wide applications. In this talk, I will introduce our pioneer work on topic-driven Chinese discourse machine reading comprehension, a key project funded by NSFC AI fundamental research program.
Generally, a discourse constructs around a topic. In this project, we plan to undergo the research on topic-driven Chinese discourse machine reading comprehension systematically and deeply from various aspects, such as theoretical systems, computational modeling and platform construction. The main content of this project includes:
1) This project extends the existing rhetorical structure theory and proposes a micro and macro combined discourse rhetorical structure representation scheme.
2) This project first proposes a theme and rheme theory based discourse micro topic structure representation way, then construct micro topic chains based on theme progression modes and finally proposes a micro and macro combined discourse topic structure representation scheme.
3) This project proposes a micro and macro combined discourse rhetorical and topic structure unified representation scheme by efficiently connecting above two representation schemes through a discourse connectivity middleware. On the basis, this project proposes a set of suitable annotation guidelines and constructs a corpus of certain scale.
4) This project proposes and implements a set of theme progression theory based computational models towards discourse rhetorical and topic structure unification.
5) This project implements a platform on Chinese discourse machine reading comprehension, which provides fundamental supports for discourse-level applications, such as document summarization and question answering.

About the Speaker: Dr. Zhou Guodong is a professor (Grade II) in Soochow University, China. His research interests include artificial intelligence, natural language understanding and natural language cognition. Till now he has published over 150 highly competitive peer-reviewed papers in SCI journals and CCF A/B conferences with Google scholar citations of over 6300. Currently, he is in charge of two NSFC key projects and one NSFC normal project. He once served as the Editorial board member of Computational Linguistics and currently serves as an associate editor of ACM TALLIP, an elite reviewer of TACL, an executive editorial member of Chinese Journal of Software, a vice chairperson of CCF TCCI and a member of Soochow University Academic Committee as well as a (senior) program committee member of first tier venues, such as ACL, AAAI, IJCAI.

Return

Description: The media industry is undergoing transformation from being paper-based to digital; meanwhile, intelligent media characterized by knowledge mining, big data analysis and intelligent interaction has become the new norm in industry. This report, which roots from needs and prospect of intelligent media services, introduces research and applications in the domains of knowledge mining and analysis, media big data analysis and autonomous regeneration of media content, innovations and breakthroughs in NLP and CC technologies, and series of technological outcomes and products achieved by Founder Apabi and the State Key Laboratory for Digital Publishing Technologies both of which are subsidiaries of Founder IT.

Talk 1: Applications and Exploration of NLP & CC technologies in the field of intelligent media, by Xiaojun Huang

Abstract: In recent years, natural language processing technology and research in Chinese computing has gone more profound and wider; at the same time, intelligent media is on the fast track and provides adequate application scenarios. This report is going to explore how to apply natural language processing and Chinese computing including text mining, machine learning, knowledge graph, intelligent Q&A, knowledge mapping and formula retrieval to the numerous application scenarios of intelligent media in order to realize construction and display of professional knowledge base, full media library personalized recommendation, resource association search and discovery, library intelligent services, formula retrieval, replication and re-editing of formulas etc.

About the Speaker: Xiaojun Huang, Professor-level Senior Engineer, Senior Technology Expert, Vice-dean of the Apabi Research Institute, Academic Committee Member of the State Key Laboratory for Digital Publishing Technologies. Mr. Huang has been a member on the Chinese Information Technology committee of the CCF since 2012. He has been working in the industry of digital publishing, digital library, knowledge service etc. for many years; in addition, Mr. Huang has led a National Science and Technology Support Plan project, 2 Press and Publication Major Science and Technology projects, and has participated research activities organized by the Electronic Information Industry Foundation, the National Development Committee Information Security Special Project, the Innovative Industry Development Special Project of the City of Beijing, National Science and Technology Support Plan etc. As a crucial member Huang has taken part in projects which have been rewarded 2nd prize of National Science and Technology Progress Award; One of Mr. Huang’s notable works, the Chinese Digital Book Garden, has been gifted to clients abroad including the Cambridge University by government officials.

Talk 2: Application and Innovation of CNDPLAB in the Field of Knowledge Service, by Haihua Xie

Abstract: Knowledge service is based on collection and organization of knowledge and aims for providing clients with accurate and professional knowledge-related services. Fundamental tasks of knowledge service involve retrieval, organizing, analysis and regrouping of knowledge, which are supported by NLP technology. Our team have achieved significant innovations and technical breakthroughs in basic NLP tasks including sentence segmentation, information extraction, text classification, and advanced tasks such as autonomous Q&A and comprehensive reading. These innovations and breakthroughs have been applied to analyses of legal documents and news, building platform for academic big data and intelligent library system, etc., and the outcomes have been overwhelmingly positive.

About the Speaker: Haihua Xie is a senior researcher in the State Key Laboratory of Digital Publishing at Peking University Founder IT group. Xie has multiple-year experience in research in the area of natural language processing and machine learning, and his research interest includes domain knowledge hierarchy construction, knowledge base construction and verification, analysis of textual resources in media and publishing. His research outcome has been applied in several products such as China Digital Library, Learning and searching system, and the cloud platform of Chinese costume culture, etc. He currently leads a research team which mainly works on analyzing the text resources of publishing and media based on techniques of natural language processing and big data. Xie received a PhD of Computer Science at Iowa State University. He has been selected for Overseas High-level Talents Introduction project of Beijing.

Talk 3: KB-QA: Recent Progress of Technologies and Application, by Ming Zhou

Abstract: KB-QA (Knowledge-based Question-Answering) utilizes the knowlege base to conduct query parsing and inference answers. With the support of the availability of large scale knowledge base, and the powrful capability of deep neural network models, researchers are making new brakthroughs in KB-QA, among many other NLP tasks. This talk will give an introduction to the recent progress of the technologies of KB-QA, including semantic parser of question queries, answer extraction and inference and ranking. We will also report our new progress in the application of KB-QA in chatbot, customer support and search engine.

About the Speaker: PhD, Vice Dean of Microsoft Research Asia, Candidate for President of ACL, Director of the Chinese Computer Society, Director of the Chinese Information Technology Professional Committee, Director of the Terminology Working Committee, Executive Director of the Chinese Information Academic Committee and Phd Supervisor of numerous universities including Harbin Institute of Technology, University of Tanjin, Nankai University, University of Shandong, etc.
Dr. Zhou is the innovator of the Chinese-English translation system CEMT-I (Harbin Institute of Tech., 1989) and the renowned Chinese-Japanese translation machine “J-Beijing” (Nippon Takasya, 1998).
Dr. Zhou joined MS Research Asia in 1999 and has been overseeing the NLP research team. He was responsible for products including the MS Input, Bing Dictionary, MS Chinese-English Translation, MS Chinese Cultural Series, etc. and made major contributions to MS Office, Bing Search, and NLP components in multiple Windows products. In recent years, Dr. Zhou has been leading his team and working with the MS Product team on chatbot systems including Xiaobing (CHN), Rinna (JAP) and Zo (US).
Dr. Zhou has published more than 120 papers in numerous conferences and journals (including more than 50 ACL articles) and is the author of more than 40 patents.

Talk 4: New Progress in Machine Writing, by Xiaojun Wan

Abstract: Machine writing is committed to enabling machines to autonomously write various kinds of high-quality manuscripts such as news, reviews, abstracts, articles and poems. Machine writing is among the most cutting-edge technologies in the fields of artificial intelligence and natural language processing, and one of the revolutionary technical advancements in publishing and media industry. This report will introduce latest progress in machine learning made by our peers in the industry and us.

About the Speaker: Xiaojun Wan, researcher in the Institute of Computer Science and Technology at Peking University, PhD supervisor, head of Language Computing and Internet Mining Research Group. Dr. Wan is an editor of Computational Linguistics, a top-tier international journal, and the standing reviewer of TACL. He serves as the area chair or senior program member for many first-class and important international conferences in the fields of Artificial Intelligence and Natural Language Processing, including ACL, NAACL, EMNLP, IJCAI, AAAI and IJCNLP. Dr. Wan receives a number of honors or awards, such as Beijing Nova Program, The MOE Program for New Century Excellent Talents in University (NCET), CAAI WUWENJUN Technical Innovation Award, ACL Outstanding Long Paper, and CCF NLPCC Distinguished Young Scientist.

Talk 5: Semantic knowledge discovery system which is based on hybrid of knowledge organization system and deep learning, by Tan Sun

Abstract: This section mainly introduces exploration and practice in improving quality and efficiency of accurate discovery, association discovery and exploration discovery by mixing knowledge organization system and deep learning of text big data.

About the Speaker: Library Scientist, PhD in Management, PhD Supervisor. Dr Sun graduated from the Department of Library Intelligence at the University of Heilongjiang with a bachelor’s degree in management in 1991; he received both of his master’s degree and PhD in management from the Centre of Document and Information of the Chinese Academy in 1997 and 2000 respectively. He currently serves as the director and deputy secretary of the party committee at the Institute of Agricultural Information, Chinese Academy of Agricultural Sciences. Dr. Sun participated in 10 national and ministerial-level research projects, published 85 academic papers on professional journals and authored and co-authored 5 books. Dr. Sun also undertook and presided over include the "Special Object Metadata" of the major science and technology infrastructure project subproject of the Ministry of Science and Technology, the "National Standard for the Application of Metadata and its Descriptive Data Standards" of the National Science Digital Library Project; moreover, he led research and construction work of the joint directory database and other national scientific digital library projects, such as the Chinese Science Citation Database and the Chinese Scientific Literature Basic Database.

Return

Description: Along with the increasing requirements of mobile users, open domain question answering has received considerable attentions in both academic and industry communities. If search engine can provide exact answers rather than related web pages for questions in natural language, users’ requirements would be better satisfied. Such challenges have been addressed from different perspectives, including Question Answering over Knowledge Bases, Open Domain Question Answering, Community based Question Answering. In this workshop, we invite researchers to introduce their recent works and discuss important research topics in this area.

Talk 1: Semantic Parsing for Search Engine & Conversational AI: Background, Methodology and Latest Progress, by Nan DUAN

Abstract: Nowadays, search engines (e.g., Google, Bing and Baidu) and conversational AIs (e.g., Siri, Alexa, Cortana/Xiaoice and Google Assistant) are changing the ways how people find information and communicate with physical devices, and improving productivities to search, e-commerce, social networks, and more. One of the core technologies behind these two types of systems is called semantic parsing, whose objective is to transform a natural language utterance into a machine executable meaning representation.
In this talk, we will first give a simple review to semantic parsing, and then introduce state-of-the-art methods for two types of semantic parsing technologies: (1) context-independent semantic parsing, which deals with single-turn questions, and (2) context-dependent semantic parsing, which deals with multi-turn questions. A list of commonly used and latest released semantic parsing datasets will be described and compared as well. Last, we will summarize the talk and point out several future directions.

About the Speaker: Dr. Nan Duan is a Lead Researcher in Natural Language Computing group at Microsoft Research Asia. He is working on fundamental NLP tasks, including question answering & generation, semantic parsing, dialogue system, paraphrasing, and etc. Many of his research have already transferred to important Microsoft AI products, such as Bing, Xiaoice, and Cortana. From 2015, he is the organizer of NLPCC Open Domain QA Shared Task.

Talk 2: Towards Building More Intelligent Chatbots, by Minlie HUANG

Abstract: Building open-domain, open-topic chatbots is one of the most challenging AI tasks due to the difficulties of natural language understanding and the requirements of world knowledge and even semantic reasoning. In this talk, I will present some recent studies towards building more intelligent chatbots: how a chatbot can express emotions via textual response, how to assign personality/profile to a chatbot, how to use commonsense knowledge to facilitate language understanding and generation, and how to ask good questions by a chatbot. I believe that these attempts will move forward to more intelligent, human-like chatting machines.

About the Speaker: Dr. Minlie Huang now is an associate professor of Dept. of Computer Science and Technology, Tsinghua University. He received his PhD degree in 2006. He was awarded “Tsinghua Excellent Researcher Fellowship” in 2006 and was selected by “Beijing Century Young Elite Program” in 2013. His work on Emotional Chatting Machine was reported by MIT Technology Review, the Guardian, NVIDIA, Xinhua News Agency, etc. He has two papers voted as Top 15 NLP papers in 2016 and Top 10 NLP papers in 2017 by PaperWeekly respectively. He has published 60+ papers on top conferences such as ACL, AAAI, IJCAI, EMNLP, KDD, ICDM, CIKM, and highly-impacted journals like TOIS, Bioinformatics, JAMIA etc. He served as area chairs for ACL 2016, EMNLP 2014, EMNLP 2011, and IJCNLP 2017, and Senior PC of IJCAI 2017 and IJCAI 2018, and reviewers for ACL, IJCAI, AAAI, EACL, COLING, EMNLP, NAACL, CIKM, ICDM, SDM, and reviewers for journals such as TOIS, TKDE, TALIP, etc. As principle investigator, he had established many collaborations with industrial companies such as Samsung, Microsoft, HP, Fujitsu, Google, ExxonMobil, Sogou, and Tencent. His homepage is at: http://coai.cs.tsinghua.edu.cn/hml/.

Talk 3: 人工智能在机对话系统中的技术现状与挑战, by Rui Yan

Abstract: 近年来，自动人机对话系统在学术界和工业界都获得了相当大的关注度，随着微软小冰百度度秘等产品的发布，以及大量人工智能公司的建立，这些对话系统背后的技术在逐步积累，也逐步解密。随着研究者的探索愈发深入，人们看到了一个对话系统逐渐从科幻电影中走进现实生活的可能。在讲座中，我将回顾人机对话的发展历程，以及随着深度学习技术盛行之后，由数据驱动模型带来的革命性改变。讲座将从人机对话的已有应用出发，再分析现有对话系统的不足，展望下一代人机对话系统的挑战。同时会分享我们组在人机对话研究所做的努力与探索，以及相应的代表性成果。

About the Speaker: Rui Yan is now a tenure-track assistant professor and PhD advisor affiliated with Institute of Computer Science and Technology (ICST) at Peking University. In the meanwhile, he is also a researcher at Beijing Institute of Big Data Research, an adjunct professor and external advisor at Central China Normal University and Central University of Finance and Economics. Previously, he was a senior researcher at Baidu Inc.. Till now he has published more than 60 papers, including papers appeared at top tier conferences about NLP/AI/IR/DM, such as ACL, AAAI, IJCAI, WWW, SIGIR, and KDD. He was invited to give a tutorial at EMNLP, and served as a (senior) program committee for important conferences such as KDD (SPC), IJCAI (SPC), AAAI (SPC), ACL (PC), SIGIR (PC), etc. His research interests are dialogue systems, information retrieval, text mining, understanding and generation.

Return