IEEE Latin America Transactions | 2021
DELFOS: A Model for Multitemporal Analysis based on Contexts History
Abstract
This paper presents Delfos, a model for multi-temporal analysis based on historical contexts. The article describes the model architecture, implementation, and tests with two real datasets to evaluate the functionality and its potential to real applications. The first scenario used historical data from an electric power transmission substation, and the second with data from the performance analysis of futsal athletes. The results obtained through inference rules perform with 100% accuracy of the Profiles creation and that these are a better representation of the present, than only the current context data of the application. The results on the prediction of contexts were relevant in the case of the existence of behavior patterns in the contexts of the analyzed entities. For the application related to the electric power transmission substation, the assertiveness was between 86% and 92% in a 12-hour interval. In the application of performance analysis of futsal athletes, we made predictions about the athletes movement, with results varying between 72% and 92% for the time of up to 20 seconds. The results were encouraging and show potential for implementing Delfos in real-life situations.