The document proposes a Knowledge Based Trust (KBT) algorithm as an alternative to existing link-based algorithms for determining page rankings. KBT would extract triples of information (subject, predicate, object) from web pages to determine relevance. It would create a "knowledge vault" to store triples and use them to calculate page rank factors through a mathematical model involving two sets of variables, theta and z, representing the probability a source provides true triples. The trustworthiness of each page would influence how much its vote contributes to determining the truth of triples, in an iterative process. This knowledge-based approach could overcome flaws in current link-based methods by privileging strong facts over mere links.
2. What to expect
1. History
2. What is KBT
3. how will it work
4. related math
5. Conclusion
6. References
3. History
Google created an algorithm to give page
ranking to different websites.
Modifications happened with many supported
algorithms.
Link based modal got many flaws.
Different attributes were created to support
the main algorithm.
KBT is proposed.
4. What is KBT
Knowledge is everything.
Strong and actual data would win ultimately.
Strong facts would get preference over links.
5. How will it work
Google will create the knowledge vault.
Triples will be extracted from each page.
These triples will play most important role in algorithm.
These will do all the math to define the page rank factor by kbt.
These triples are : subject, predicate, object.
6. Math behind_1
1. It updates two series of data theta and z.
2. Where, z=(v , c).
3. V is Any given triple is either true or false, and the c is question of
whether any given web source provides a given triple is either true
or false.
4. Theta = (theta1, theta2).
5. theta_1 is a series of variables representing the probability that a
given web source provides a “true” triple.
6. Theta2 is:
8. Conclusion
The trustworthiness of each page
influences how big its overall vote is in
determining the “truthfulness” of each
possible triple, which in turn influences the
other factors, and so on.
9. Reference
J. Bleiholder and F. Naumann. Data fusion. ACM Computing Surveys, 41(1):1–41,
2008.
K. Bollacker, C. Evans, P. Paritosh, T. Sturge, and J. Taylor. Freebase: a collaboratively
created graph database for structuring human knowledge. In SIGMOD, pages 1247–
1250, 2008.
A. Borodin, G. Roberts, J. Rosenthal, and P. Tsaparas. Link analysis ranking:
algorithms, theory, and experiments. TOIT, 5:231–297, 2005.
S. Brin and L. Page. The anatomy of a large-scale hypertextual Web search engine.
Computer Networks and ISDN Systems, 30(1–7):107–117, 1998.
Northcut.com
Xin Luna Dong, Kevin Murphy’s ‘Knowledge based trust’.