Text Mining of Open Source

Patrice Slert, Johns Hopkins University, is speaking on text mining open source.
 
By open source,she means structured data. She mines WebofScience, Dialog. It’s not drivers licenses, other public record sources.
 
She warns us to beware of data conclusions. How are a rooster crowing and the sun rising related? She notes that open source has grown in importance for the intelligence community. More emphasis on including librarians in the OSI community.
 
ISI WoK citation searching – ISI analysis tools let you visualize the numbers of publications by country on a particular topic, visualize by publications by institution.
 
Better she thinks is VantagePoint, which provides an import wizard tailored to various databases and lets you combine databases. INSPEC in VantagePoint has a details window feature, the citation field can be changed or additional windows added. There’s a prediction box based on algorithms. It’s got a thesaurus. There are author maps. Each node is clickable and will bring up the full citations. It interacts with Excel. Part of the author map is linkages so you know who is collaborating with whom. You can see who "lynchpin" person is. You can create a co-occurrence matrix to see how different fields relate to each other.
 
Silobreaker lets you mine the news. It’s very big in the intelligence community. You can do word searches, visual searches, geographic searches. Lets you find networks of people, organizations, companies, industries, key phrases. She’s demo-ing Silobreaker.
Marydee Ojala
 
 
 

Comments are closed.