Characterizing Pure High-Order Entanglements in Lexical Semantic Spaces via Information Geometry

  • Yuexian Hou
  • Dawei Song
Conference paper

DOI: 10.1007/978-3-642-00834-4_20

Part of the Lecture Notes in Computer Science book series (LNCS, volume 5494)
Cite this paper as:
Hou Y., Song D. (2009) Characterizing Pure High-Order Entanglements in Lexical Semantic Spaces via Information Geometry. In: Bruza P., Sofge D., Lawless W., van Rijsbergen K., Klusch M. (eds) Quantum Interaction. QI 2009. Lecture Notes in Computer Science, vol 5494. Springer, Berlin, Heidelberg


An emerging topic in Quantuam Interaction is the use of lexical semantic spaces, as Hilbert spaces, to capture the meaning of words. There has been some initial evidence that the phenomenon of quantum entanglement exists in a semantic space and can potentially play a crucial role in determining the embeded semantics. In this paper, we propose to consider pure high-order entanglements that cannot be reduced to the compositional effect of lower-order ones, as an indicator of high-level semantic entities. To characterize the intrinsic order of entanglements and distinguish pure high-order entanglements from lower-order ones, we develop a set of methods in the framework of Information Geometry. Based on the developed methods, we propose an expanded vector space model that involves context-sensitive high-order information and aims at characterizing high-level retrieval contexts. Some initial ideas on applying the proposed methods in query expansion and text classification are also presented.


Information geometry Pure high-order entanglement Semantic emergence Extended vector model 


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

Copyright information

© Springer-Verlag Berlin Heidelberg 2009

Authors and Affiliations

  • Yuexian Hou
    • 1
  • Dawei Song
    • 2
  1. 1.School of Computer Sci. & Tech.Tianjin UniversityChina
  2. 2.School of ComputingThe Robert Gordon UniversityUnited Kingdom

Personalised recommendations