Rating-Based Guidance System for Public Safety Using Classified Localities

Abstract

To automate the manual SOS procedure given in Kavalan mobile application and to provide users a safe path by the given source and destination (from the users). This work uses regression algorithms such as linear regression, decision trees, or support vector machines (SVM) to predict the rating for a current zone which can be used as input for generating the graph with rating as weights and the graph is used as the input for the Dijkstra algorithm which produces the safest path based on the rating. Thus, this path can be used to navigate the public safely to their destination while avoiding unsafe zones. Furthermore, a feedback form is available using which the user can provide textual as well as numerical feedback regarding the places they travel. Decision tree regression provides an accuracy of 89.4% compared with the other regression models since the dataset is categorical and less in size. The safety path is also being produced using Dijkstra's algorithm and the feedback is analysed using the T5 model.