In this paper, a hybrid approach for finding safe routes using semantic processing and classification algorithms, with data provided from a social network and the official crime reports, is presented. As a case study, a mobile information system was developed. It generates safe routes based on crime reports of Mexico City from large tweet repository and official databases. The data are semantically classified to determine whether the tweet describes a crime event or theft; in case of tweets which cannot be identified as crimes, they are evaluated by Bayes algorithm, which clustered them according to the contained description. Thus, the clusters are used to make prediction regarding the possibility that a crime can occur in a specific place and hour. The spatiotemporal analysis determined the location where the crime events occurred. Moreover, the confidence level of a location was defined and it was used as a parameter for computing the safer route. The main contributions of this work are as follows: (1) the design of a hybrid approach based on semantic processing to retrieve crime data from a social network source; (2) the integration of crowd-sensed data with official government sources; (3) the validation of a tweet performed by comparing the sources, using -fold cross validation; (4) the estimation model based on the Bayes algorithm to obtain safe routes with data that were provided by the mobile device; and (5) the design of a mobile information system to generate safe routes. According to the results of the estimation. the certainty degree is around 75% of effectiveness. It was tested by comparing areas with crime data, but the records were intentionally removed and original copy was kept. Thus, with the results of the estimation, a comparison with the original copy was performed. So, we found that the estimation has a performance of 75% for all the points of the data sample. In addition, a metric to measure the confidence level or security for certain points and areas of Mexico City has been proposed. It allows finding safe routes, according to paths with a low crime rate. Moreover, the mobile application gathers long-term statistical data with almost real information from citizens, which are acting as sensors in the city. The results of the mobile system have been tested and compared with the Crime Map System.
|Country:||United States of America|
|Institution:||Mobile Information Systems |
|Author:||FélixMata, Miguel Torres-Ruiz, Giovanni Guzmán, Rolando Quintero, Roberto Zagal-Flores,MarcoMoreno-Ibarra, and Eduardo Loza|