Content quality has been a longstanding problem for information access. With the rapid progress of news feed in recent years, the importance of hight quality content cannot be overstated. In this talk, I will share Baidu’s practice in developing a leading news feed product, with a special focus on building content quality models with AI technology.

Dr. Yanjun Ma is a scientist at Baidu, with research mainly focusing on Natural Language Processing and Machine Translation. His research is widely used in a range of products including Baidu web search, news feed and Duer. He has authored and co-authored over 20 research publications, and served on the editorial board of Machine Translation Journal, as well as area co-chairs for a number of international conferences including ACL, IJCAI etc. Dr. Yanjun Ma obtained his PhD in Computer Science from Dublin City University in 2009. He also holds MA and BA in Computational Linguistics from Tsinghua University and Shandong University. In 2015, Dr. Yanjun Ma received 2015 China Science and Technology Progress Award for his research on Machine Translation.

Title: Four Waves of AI Business
            – NEC the WISE and NEXT –

Up to now, we are experiencing three waves of AI utilization in the real business world. The first one was the utilization of Recognition by AI, the second one was the utilization of Prediction by AI, and the third one is the utilization of Control by AI. NEC the WISE is a portfolio of these AI technologies developed by NEC for enriching human intellect and creativity.
After quick overview of “NEC the WISE”, I will introduce one of our machine learning engine “Heterogeneous Mixture Learning”, which outputs “White-Box” type predictor based on training data. In real business, interpretability often becomes crucial to satisfy customers, and this engine has been a main solution in our business.
Also, I will mention about the approaching fourth wave, Negotiation. Coordination among AI systems will be necessary in the near future, and automatic negotiation technology seems one of the keys to that.

He has been working on R&D of AI technologies and business development of them. Recently, his main activities are on R&D of automatic negotiation and automatic reasoning. He received his Ph.D. degree in mathematical engineering from the University of Tokyo in 1999.

Title: Data as gravity – Yahoo! JAPAN’s collaborative challenges

Understanding users – their interests, preferences, and context -, is the most important task for internet businesses. Yahoo! JAPAN is in quite a unique position from this point of view. It has multiple services such as web search, e-commerce, news, Q&A, and credit cards with a good amount of users, which will bring various touch points, more opportunities to create values, as well as multi-angled user behavior data.
In this talk, I will share Yahoo! JAPAN’s challenges to leverage this “multi-big data” ranging from organizational aspects to technical contributions.

As Head of Yahoo! JAPAN Research, Akira Tajima is working on leveraging advanced technologies and big data in Yahoo! JAPAN services, as well as leading research activities. Prior to joining Yahoo Japan Corporation, he led Mathematical Science team in IBM Research – Tokyo. He also worked as a consultant at A.T. Kearney for several years and enhanced his business background. He received his Ph.D. degree in information science from the University of Tokyo in 2000.

In this talk, we will start from a most sucessful machine learning system, known as Baidu’s Phoenix Net system, in China. We will show the thinkings of the design of this very high dimensional machine learning system, from the version 1.0 in 2009 to version 2.0 in 2013. This system helped Baidu boost its revenue for over 800%. Then we will show the main difficulties if we want to apply such kind of technologies to more other applications, and introduce some solutions we are working on to overcome these difficulties.

Wenyuan Dai is the founder and CEO of the 4Paradigm Tech Co., Ltd. 4Paradigm is a leading AI service provider mainly for financial and internet enterprises. Before starting 4Paradigm, Wenyuan was the senior scientist and principal architect in Baidu, and responsible for Baidu’s search ads machine learning system. Wenyuan was also a principal research scientist of Huawei’s Noah’s Ark Lab, and led a machine learning research team covering Huawei’s business in telco, finance, and mobile phone.

Title: Textual Information based automatic Item Classification
            – a case of Challenging Company in Japan –

EC companies today handle millions of items to meet more and more diverse demand of the consumers. These items must be classified into hierarchical categories (sometimes by hand) in order to make it easier for consumers to find what they want, and to make it easier for companies to manage items.
In this presentation, a case in a Japanese EC company will be introduced. I will talk about our challenge to classify these items automatically using textual information about items, and I will report our results including fastText.

Kei Harada was received Ph.D. degrees in Mathematical Science from Nagoya University, Japan in 2011. He is working as a consultant and as a data scientist in Financial Engineering Group. He is a Kaggle Master, and his team won the 2nd prize in KDDCUP 2015.