High Power Computational Analytics for Media Assessment
Abstract
Section I. Abstract This DURIP proposal is to improve the infrastructure for carrying out research on advanced analytical techniques for analyzing news media outlets, social media channels and remote sensor data with the goal of identifying and tracking a number of important military objectives. The proposed coarse-grain processor cluster and massive data storage system will give us the capability to carry out research on new algorithms for carrying out network analysis on mixed over-time data sources (e.g., social media sources, new sources, and remote sensor data) in order to extract understanding and actionable intelligence. The need for a dedicated coarse-grainparallel processing cluster with a very large data store is driven by types of data being analyzed; e.g., over 20 million tweets, over 100,000 news articles. In addition, researchers will carry out simulations of social-cyber attacks, such that the models are instantiated with data from the news and / or twitter and then the processor array is used to forecast change in these data under alternative conditions. Researchers will also store intermediate and final results from machine learning studies used to classify data in the twitter and news into topic categories. Also, researchers will use this proposed system for running agent-based simulations of social network evolution and topic network evolution. Finally, researchers will develop algorithms for performing big-data analytics, including testing new network analysis algorithms, on the real and simulated data. This will include studies of parallel processing algorithms for media data analysis to evaluate the asymptotic efficiency of using additional processors as a function of the algorithm. In summary, the proposed system will greatly expand our ability to perform research and teaching in this area. In addition, the proposed hardware will allow us to train a new generation of graduate students in this challenging research area.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Aug 08, 2016
- Source ID
- N000141512869
Entities
People
- Kathleen Carley
Organizations
- Massachusetts Institute of Technology
- Office of Naval Research
- United States Navy