Semantic information valuation
Gregory L. Heileman, Sinan al-Saffar · 2009
Information quality is implicitly utilized and mostly non-subjectively computed in information retrieval (IR) systems. We discuss the disadvantages of this approach and instead explicitly define and compute the value of an information piece as the potential semantic impact this information can subjectively have on a recipient's world-knowledge. We model the problem using two methods: semantic impact graphs and RDF graph dissimilarity. We compute information value using these graph models. While the former may present a challenge to software agents in tracing information provenance to infer trust, the latter automates that computation through the semantic overlap in RDF graphs. Two graphs are constructed, one representing semantics of the target body of information and one representing the context of the consumer of that information. We then compute information value as change to the context graph resulting from the information-consumer learning of the target graph. This change is computed as a potential graph edit distance measuring the dissimilarity between the context graph before and after the learning of the target graph. A particular application of this subjective information valuation is in the construction of a personalized ranking component in Web search engines or more generally as a heuristic to guide the search direction in knowledge-based expert systems. Based on our methods, we construct a Web re-ranking system that personalizes the information experience for the information-consumer.