Research Website

http://www.purdue.edu/discoverypark/vaccine/

Keywords

Visual Analytics, VAST Challenge, Email Networks, News Article Analysis

Presentation Type

Event

Research Abstract

News articles and email conversation data could be very useful in the analysis of developing and ongoing events, such as preventing a potential threat or possibly even locating a missing person. There is currently no “one-size-fits-all” solution to visualizing diverse forms of datasets and their sheer sizes are far too great to efficiently analyze by brute force methods. However, using principles of Visual Analytics, it is possible to take this information overload and transform it into a useful tool to help increase the efficiency of event analysis. A visualization system was developed for email conversation networks using web technologies. An interactive force diagram was constructed, allowing for an easy analysis of communication links between people. This force diagram was able to be filtered down to specific people or emails and with color coded nodes based on positions held in a company. A dynamic list of email headers was created that allowed for filtering based on specifically chosen people or by user defined importance. Lastly, a slide-out menu was implemented to allow for a side by side comparison between two selected people by displaying their employee records. The system created was used on a data set from the VAST 2014 mini challenge 1 and it allowed for the successful analysis of a fictional companies email network. Although this specific system was designed around the VAST 2014 data set, it could easily be modified to work with diverse email conversation network data to aid in various forms of analysis.

Session Track

Data Analytics

Share

COinS
 
Aug 7th, 12:00 AM

VAST 2014, Challenge One: Event Analysis Within Big Data

News articles and email conversation data could be very useful in the analysis of developing and ongoing events, such as preventing a potential threat or possibly even locating a missing person. There is currently no “one-size-fits-all” solution to visualizing diverse forms of datasets and their sheer sizes are far too great to efficiently analyze by brute force methods. However, using principles of Visual Analytics, it is possible to take this information overload and transform it into a useful tool to help increase the efficiency of event analysis. A visualization system was developed for email conversation networks using web technologies. An interactive force diagram was constructed, allowing for an easy analysis of communication links between people. This force diagram was able to be filtered down to specific people or emails and with color coded nodes based on positions held in a company. A dynamic list of email headers was created that allowed for filtering based on specifically chosen people or by user defined importance. Lastly, a slide-out menu was implemented to allow for a side by side comparison between two selected people by displaying their employee records. The system created was used on a data set from the VAST 2014 mini challenge 1 and it allowed for the successful analysis of a fictional companies email network. Although this specific system was designed around the VAST 2014 data set, it could easily be modified to work with diverse email conversation network data to aid in various forms of analysis.

http://docs.lib.purdue.edu/surf/2014/presentations/141