Research Intelligence

The Research Intelligence project aims to provide a better understanding of the scientific research conducted by principal investigators associated with the California Institute for Telecommunications and Information Technology (Calit2) and to match these strengths to a deeper understanding of federally sponsored research funding.

Our project proposes to develop novel techniques for organizing and visualizing organizational links, using recently developed methods that are at the intersection of state-of-the-art research in text mining and social networks. The algorithms and software will allow users to quickly visualize the topical content of a document collection, and see connections between the documents' authors. With our work, users will be able to easily examine the scope of Calit2 research. Furthermore, our models will allow Calit2 researchers to find other Calit2 researchers with similar interests, and automatically be alerted to relevant fun opportunities. In the broader sense, extracting links and relationships from text data, and using this inferred information as the basis for interactive querying and visualization, is a concept that has broad applications in academia, government, and industry

It should be noted that all data and interfaces demonstrated are in "alpha" iteration and will be enhanced as the project continues.

Project Team Members:

• Padhraic Smyth, Professor, Department of Computer Science, UC Irvine
• *Mohan Paturi, Professor, Department of Computer Science, UCSD
• Jerry Sheehan, Manager Government Programs, Calit2, UCSD
• David Newman, Project Scientist, Department of Computer Science, UC Irvine

*Dr. Paturi has acted as a technical advisor and will be a future collaborator on UCSD/Calit2 data mining initiatives.

Topic Modeling
The topic model uses statistical learning algorithms to automatically discover the topics that describe a document collection. A "topic" is a probability distribution over words, with high probabilities assigned to words that are associated with that topic. Given a set of topics, we can use the probability model to automatically characterize the nature of a researcher's work and find other researchers with similar interests.
Link to topic browser

Data Funding Visualization
The Calit2 project team has assembled comprehensive data on research funding at UCSD from the public and private sector since the initiation of the Institute. Using data visualization tools developed by the University of Maryland for creating "tree maps" the team has created a dynamic environment for exploring research funding interactively.
Link to tree map applet