Towards a New Method for Extracting Uncertain Data

The imperfections in evidential data are modeled using evidence theory. While there are several algorithms for extracting logical data available in the literature, there has been limited research on frequent evidential items, especially within tree-based evidential data structures. This paper presents the problem of extracting frequent itemsets from uncertain data, especially the evidential data. To tackle this issue, we introduce a novel approach called EEvFIT (Extraction of Evidential Frequents items based on Tree structure), This method is specifically designed to extract frequent items using a tree structure, addressing the inadequacies of existing techniques. The adopted technique generates a good score of frequent items during the first tree run, and it increases significantly during the second one on the same tree. EEvFIT has been tested on both UCI evidential and synthetic datasets by taking into consideration the case of dense and sparse databases. The tests showed the effectiveness of our proposed EEvFIT method according to several parameters such as extraction time and number of frequent items compared to existing methods. This demonstrates that the new approach is not only necessary but also provides substantial improvements over current methods.