On the Equivalence of Semantic Networks and Vector Spaces under Ranked Retrieval

Vladimir Kulyukin and Amber Settle
School of Computer Science, Telecommunications,
and Information Systems
DePaul University
243 South Wabash Avenue
Chicago, IL 60604
vkulyukin@cs.depaul.edu, asettle@cs.depaul.edu

Abstract:

A formalism is presented for ranked retrieval from a universe of objects. Two retrieval models, semantic networks with spreading activation and vector spaces with dot product, are formalized and proved equivalent under ranked retrieval.

• Introduction
• Definitions
• Retrieval models
  • Semantic networks
  • Vector spaces
• Equivalence of models
  • From vector spaces to semantic networks
  • From semantic networks to vector spaces
• Conclusion
• Appendix
  • Listing 0
  • Listing 1
• References
• About this document ...