Practicing the Art of Data Science

Data science embraces interdisciplinary methodologies and tools, such as those in statistics, artificial intelligence/machine learning, data management, algorithms, and computation. Practicing data science to empower innovative applications, however, remains an art due to many factors beyond technology, such as sophistication of application scenarios, business demands, and the central role of human being in the loop. The purpose of this keynote speech is to share with the audience two most important rules of thumb that I learned from my practice of data science research, development and applications, as well as my thoughts on the future enterprise and organization data strategies. First, I will demonstrate the importance and challenges in developing domain-oriented, end-to-end solutions. Specifically, I will discuss our experience in transforming algorithms to domain-oriented tools, and review some of our latest techniques in transforming black-box deep learning networks into interpretable white-box models. Second, I will advocate the core value of data science as the connector and transformer between vertical application challenges and general scientific principles and engineering tools. Using network embedding as an example, I will illustrate the innovative value of building connectors and transformers for new types of data and applications so that they can take great advantage of well established scientific methods and engineering tools. I envision data science for social, commercial and ecological good has to build on enterprise and organization data strategies and infrastructure. About future work, I will provide some thoughts on this perspective, such as data value assessing and pricing, as well as privacy preservation.