Summary
In this talk, Professor Ding gives an inspiring talk about data
science’s present situation and future. She points out that with the
advancement of technology, the production and access to large-scale data sets have
fundamentally changed how people think and how they live. At the same time, it
also brings lots of valuable opportunities and challenges for data scientists.
Researchers coming from different fields might focus on
different levels of problems. Methodologists coming from mathematics and physics
tend to look at things at a macro-level, trying to extract some rules or build
general models to explain the complicated, dynamic, stochastic and messy data
patterns. Some researchers pay much attention to issues at micro-level, such as
in recommendation system, they investigate each individual’s behaviors and preferences.
While, data scientists prefer to resolve problems arising from meso-level, playing
the role of intermediators connecting micro and macro levels.
Specifically, Prof. Ding presents some recent work and obtained results
by her research group, mainly about bibliometrics (or science of science), drug
protein target interactions, and innovation diffusion. Bibliometrics is
statistical analysis of digital publications, such as papers, articles, books
and news. For example, given an academic publication corpus, they can mine
knowledge with respect to collaboration, citation as well as research topic
shifts over time etc. Besides, her research group is also interested in predicting
drug protein interactions based on semantic network analysis. The semantic
network integrates chemical, pharmacological, genomic, biological, functional,
and biomedical information, in which nodes and edges are extremely
heterogeneous. Due to its heterogeneity, they examine meta-path-based
topological patterns to predict potential drug-protein links. Details can be
seen in their recent publication entitled “Predicting drug protein target
interactions using meta-path based semantic network analysis”. In addition,
Prof. Ding also talks about their work on innovation diffusion. They mine a
whole collection of publications during the past decade, aiming to check out
how LDA algorithm diffuses across diverse domains. They find that the early adopters
of one innovation are responsible for the final spreading scale. It gives
researchers some good insights of how to better spread their ideas and work.
Details About Talk
Title: Data-Driven Science of Science
Speaker: Ying Ding
Bio: Dr. Ying Ding is an Associate Professor at School of
Informatics and Computing and is currently associate director of data science
online program at Indiana University. She has been involved in various NIH, NSF
and European-Union funded projects. She has published 190+ papers in journals,
conferences, and workshops, and served as the program committee member for 180+
international conferences. Her current research interests include data-driven
knowledge discovery, Semantic Web, knowledge graph, scientific collaboration,
and the application of Web Technology.
No comments:
Post a Comment