Micro level data aggregation based on profiling and Jaro-Winkler distance maximization

This article explores a micro-level data aggregation method based on feature profiling and Jaro-Winkler distance maximization. The goal is to develop an efficient approach for data aggregation that takes into account the similarity of rows and provides high accuracy of results without using the ID of the objects under study. This approach preserves confidentiality and solves the problem of mismatching a set of data objects. As part of the study, a method based on the task of assigning and searching for similar profiles was proposed, which allows you to optimally match strings and perform aggregation. The results of experiments were presented, which confirm the effectiveness. Further development of the study may include the use of a multi-agent approach to reduce the dimension of the assignment problem.