Message Importance Measure and Its Application to Minority Subset Detection in Big Data
Pingyi Fan, Yunquan Dong, Jiaxun Lu, Shanyun Liu · 2016
Message importance measure (MIM) is an important index to describe the message importance in the scenario of big data. Similar to the Shannon Entropy and Renyi Entropy, MIM is proposed to characterize the uncertainty of a random process and some related statistical characteristics. Moreover, MIM also needs to highlight the importance of those events with relatively small occurring probabilities, thereby is especially applicable to the big data scenario. In this paper, we define a parametric MIM measure from the viewpoint of information theory and then investigate its properties. We also present a parameter selection principle that provides answers to the minority subsets detection problem in the statistical processing of big data.