Measuring the coverage and redundancy of information search services on e-commerce platforms

Today’s widespread e-commerce applications pose a new challenge to information search services. They must extract a useful small set of search or recommendation results from a larger set that preserves information diversity. This paper proposes a novel metric setting to measure two important aspects of information diversity, information coverage and information redundancy. In addition to content coverage, we consider another important measure of information coverage called structure coverage, and model it using information entropy. This approach can better convey the information coverage of the extracted small set with respect to the original large set. The proposed metrics are effective and have various useful properties, which are demonstrated by theoretical and experimental analysis. We also designed a calculation method that shows good computational efficiency. Finally, we conducted an experiment using real data from online customer reviews to further emphasize the effectiveness of the proposed metrics.

► We propose a novel metric setting to measure information coverage and information redundancy.
► We model information structure using information entropy in information coverage metric.
► The two metrics can be calculated effectively and efficiently without human intervention.
► Our proposed metrics have both higher interpretability and better properties than existing related metrics.
► The experiment using customer reviews data supports the effectiveness of the proposed metrics.