Online assessment of the value of information for searchers of a bibliographic data base

The object of the experiment described in this paper was to assess the value of information for individual searchers in a realistic search environment. An additive linear function, which combined the ratings of information features provided by searchers, was used to model how humans value information. Twenty Industrial Engineering graduate students and faculty, who were attempting to satisfy a genuine information need by searching a data base of technical literature on Operations Research, served as experimental subjects. Included in the data analysis are a comparison of two alternative retrieval methods, an evaluation of how well humans are able to estimate the relevance of a document from the document's features, and an investigation of the value function components (features and coefficients) and functional form.