REPLY TO THREAD
|
Subject
|
The Air Force wants to know what your are saying . . . Do your care?
|
User Name
|
|
|
|
|
Font color:
Font:
|
|
|
|
Original Message
|
New Metrics for Blog Mining Brian Ulicnya, Ken Baclawskia, Amy Magnusb aVIStology, Inc., 5 Mountainview Dr., Framingham, MA 01701 Air Force Office of Scientific Research, 875 North Randolph Street, Suite 325 , Arlington, VA 22203
ABSTRACT Blogs represent an important new arena for knowledge discovery in open source intelligence gathering. Bloggers are a vast network of human (and sometimes non-human) information sources monitoring important local and global events,and other blogs, for items of interest upon which they comment. Increasingly, issues erupt from the blog world and into the real world. In order to monitor blogging about important events, we must develop models and metrics that represent blogs correctly. The structure of blogs requires new techniques for evaluating such metrics as the relevance, specificity, credibility and timeliness of blog entries. Techniques that have been developed for standard information retrieval purposes (e.g. Google's PageRank) are suboptimal when applied to blogs because of their high degree of exophoricity,quotation, brevity, and rapidity of update. In this paper, we offer new metrics related for blog entry relevance,specificity, timeliness and credibility that we are implementing in a blog search and analysis tool for international blogs. This tools utilizes new blog-specific metrics and techniques for extracting the necessary information from blog entries automatically, using some shallow natural language processing techniques supported by background knowledge captured in domain-specific ontologies. Keywords: Blog mining, blog search, information fusion, relevance, specificity, timeliness, credibility. [link to www.dtic.mil]
|
Pictures (click to insert)
|
| | | | | | | | | | | | | | | | | | | | | | | | | Next Page >> |
|