Accession Number:

AD1034782

Title:

XDATA

Descriptive Note:

Technical Report,01 Sep 2012,31 Dec 2013

Corporate Author:

Raytheon BBN Technologies Columbia United States

Personal Author(s):

Report Date:

2017-05-01

Pagination or Media Count:

41.0

Abstract:

For the XDATA project we researched and implemented software concerning graph analytics and natural language processing NLP. Throughout these efforts we attempted to focus on the problems that arose when implementing the algorithms in parallel. We also partnered with subcontractors SentiMetrix and Johns Hopkins University for further graph analytics. On the graph side, we researched and provided an implementation of a graph matching algorithm, as well as a process finder which identifies patterns in graphs with a temporal element. On the NLP side, we did work on word embeddings, and provided parallel implementations of a wordcloud algorithm as well as a package that combined topic clustering with a Chow-Liu tree to estimate joint probability distributions between the topics revealed in the clustering. We worked with the Twitter, BitCoin, Akamai traceroute, and Akamai CIDR datasets.

Subject Categories:

  • Numerical Mathematics

Distribution Statement:

APPROVED FOR PUBLIC RELEASE