Methodologies and Metrics for Assessing the Strength of Relationships between Entities within Semantic Graphs
Semantic graphs are becoming a valuable tool for organizing and discovering information in an increasingly complex analysis environment. This paper investigates the use of graph topology to measure the strength of relationships in a semantic graph. These relationships are comprised of some number of distinct paths, whose length and configuration jointly characterize the strength of association. We explore these characteristics through the use of three distinct algorithms respectively based upon an electrical conductance model, Newman and Girvan's measure of betweenness [5], and cutsets. Algorithmic performance is assessed based upon a collection of partially ordered subgraphs which were constructed according to our subjective beliefs regarding strength of association.
- Research Organization:
- Lawrence Livermore National Lab. (LLNL), Livermore, CA (United States)
- Sponsoring Organization:
- USDOE
- DOE Contract Number:
- W-7405-ENG-48
- OSTI ID:
- 883760
- Report Number(s):
- UCRL-TR-216074; TRN: US200615%%221
- Country of Publication:
- United States
- Language:
- English
Similar Records
Algorithms and architectures for high performance analysis of semantic graphs.
Knowledge Representation Issues in Semantic Graphs for Relationship Detection