◆  Industrial Talk 1, by Jie Li

Topic: Application of multi­mode Human­computer interaction scenario

Date and Time: Oct. 17, 2020, 10:00­-10:20

Meeting Room: European Garden

Short Bio: Jie Li, Zonetyet VGM/CTO，Twenty years of IT industry experience, full stack technical ex­pert, early investor and CTO of Zoneyet Technol­ogy. Natural Language Processing Laboratory co­head of Zhengzhou University and Zoneyet Technology, used to be the leader of Project 985 of Shaanxi Normal University, the founder of the social welfare project ’Rui Jie Xun Zi’, now involved in the formulation of , currently responsible for the technical work of Zoneyet Technology.

Abstract: Multi­-modal human-­computer interaction integrates various forms of information input, such as body movements, language and emotion analysis, micro­-expressions, etc., so that ma­chines can feel the changes of human emotions and respond to them in a timely manner. Multimodal systems will provide a practical way and means for human­-computer interac­tion, including emotional interaction in the future. From the point of view of the history of human-­computer interaction, multi­-mode human-­computer interaction is the most ideal and natural way available currently. Multi­-modal man­-machine allows intelligent products around your life to quickly perceive your needs, and can quickly solve your needs, or pro­vide a way to solve your needs; Wearable devices, smart home, Internet of Things and other fields are gaining popularity and landing in the science and technology circle. Comprehen­sively building intelligent life will become the focus of the future industry. The multi­-mode human­-computer interaction will gradually become the key link to realize this kind of life simple interaction mode and the evolutionary application of service mode.

◆  Industrial Talk 2, by Dr. Youzheng Wu

Topic: Intelligent Human­Computer Interaction and its applications

Date and Time: Oct. 17, 2020, 11:45­-12:00

Meeting Room: European Garden

Short Bio: Dr. Youzheng Wu is an algorithm scientist of JD AI Research and Director of Intelligent Human­-Computer Department, JD Could&AI. He is leading a team whose focus is cutting­ edge technology breakthrough of multi­-turn di­alogue, natural language generation, and knowl­edge graph, and their large­-scale industrial ap­plications in customer service, marketing and sales, and digital media. Before this, Dr. Wu was a researcher in National Institute of Infor­mation and Communications Technology, Japan, University of Edinburgh, SONY China Research Lab, iQIYI, etc. Dr. Wu has published many papers in top­-tier international conferences and journals such as ACL, EMNLP, AAAI, IJCAI, Computer Speech & Language, etc.

Abstract: The mission of Human­-Computer Interaction (HCI) is to solve “communication barrier” between human and computer. Though Graphic User Interface (GUI) and Natural User In­terface (NUI) become convenient for users to communicate with computers, there still exists high “communication barrier” when billions of public users complete diverse and complex tasks such as e­-shopping, order cancellation due to the lack of understanding of scene knowl­edge. Intelligent HCI (iHCI) is expected to solve the problem. In this talk, we start with real­-word applications of iHCI with multi­-modal and multi­-turn di­alogue as its prominent characteristics in customer service, proactively marketing and sales, and digital media. We also introduce our recent progress on multi­-modal information extrac­tion, conversational machine reading compression (MRC), and multi­-modal dialogue.

◆  Industrial Talk 3, by Dr. Min Chu

Topic: Progresses and challenges in mass production of conversation bots

Date and Time: Oct. 18, 2020, 09:45-10:00

Meeting Room: European Garden

Short Bio: Dr. Min Chu serves as the vice president in AISPEECH since 2017.9. She found the R&D center in Beijing and the business unit of Intelli­gent Services. Her key responsibility is to lead the R&D in technologies and products as well as the business development of mass implemen­tation of conversation bots for various business scenarios. Dr. Chu received her PhD from the Institute of Acoustics, CAS. She worked for In­stitute of Acoustics, Intel China Research Cen­ter and Microsoft Research Asia respectively as a research faculty for more than 10 years. Her research areas include speech synthesis & recog­nition and natural language processing & appli­cations. She has published more than hundred papers in top journals and conferences and has been issued 30+ patents. Between 2009­-2017, she worked for Alibaba, leading various big data project such as machine translation, Taobao prod­uct inventory mining, user profiling, CNZZ data bank etc. She found the Intelligent Speech In­teraction team under iDST and now is under Al­ibaba DAMP Academy.

Abstract: In recent years, with the rapid progress in speech and natural language processing technolo­gies, conversation bots are considered as very important solutions for reducing cost and im­proving the qualities for customer services. When we serve the requests from different cus­tomers, we found that every customer has his own special requests. Customization for each of them is almost inescapable. Therefore, when we try to produce conversation bots as a large scale business, we face lots of challenges beyond improving the accuracies of algo­rithms. This presentation will share our insights in these challenges and the progresses we have made.

◆  Industrial Talk 4, by Dr. Haihua Xie

Topic: NLP based Knowledge Service Applications

Date and Time: Oct. 18, 2020, 10:00-10:15

Meeting Room: European Garden

Short Bio: Dr. Haihua Xie is a senior researcher in the State Key Laboratory of Digital Publishing Tech­nology. His research interests include knowl­edge base construction and verification, knowl­edge service, textual resources analysis and pro­cessing in the industry of media and publish­ing. His research achievements have been used to analyze, organize and recompose the massive data of digital publishing, and effectively man­age and utilize the digital publishing data based on knowledge systems, which significantly im­prove the traditional knowledge organization and management mode.

Abstract: In this talk, we present our research progress and achievements of knowledge service applica­tions based on Natural Language Processing technology, and how it improves the efficiency and quality of knowledge system construction in the field of digital publishing, and promotes the technological progress of digital publishing industry.

◆  Industrial Talk 5, by Dr. Fei Huang

Topic: NLP Technologies and Applications in Digital Economy

Date and Time: Oct. 18, 2020, 11:45-12:00

Meeting Room: European Garden

Short Bio: Dr. Fei Huang is senior director and principle researcher of Language Technologies Lab, Al­ibaba DAMO Academy. He leads R&D on NLP foundational technologies, dialogue and machine translation. His team develops various NLP tech­nologies ranging from lexical, syntactical, se­mantic, discourse as well as deep learning­-based algorithms, and integrate them into the Alibaba NLP platform which supports several hundreds internal and external clients with advanced NLP models, systems and solutions. Before Alibaba, Dr. Huang led the development of machine trans­lation system at Facebook AML, and was a senior researcher as IBM Watson after he gradated from CMU with Ph.D. focusing on language technolo­gies.

Abstract: The goal of natural language processing (NLP) is to enable computer to “understand” hu­man languages. In the age of digital economy, many NLP technologies are driven by data and knowledges, including linguistic data, business data as well as domain knowledges. In digi­tal economy platforms such as Alibaba ecosystem, these data and knowledges are generated, processed and utilized to train machine learning models and systems for various applications such as information extraction, title rewriting, question answer, dialogue system and transla­tion. etc.. Overall, the progress in NLP technologies, accumulation of data and knowledges as well as powerful computing platforms make it easy to do business anywhere.