Data Mining to Identify Fraud Suspected on Electronic Elections

Many democratic countries choose their representatives through electronic elections. Even being a modern tool, its results can be explored maliciously. Because that, many instruments and protocols are using to protect electronic elections from attacks. This work propose a new system to improve the security in electronic elections. It is based on analyses of behavior voter to detect urns with dissonant result, what can serve as start point to auditing. The system uses data mining to select suspect urns, presenting good results in the task of indicating urns with suspect of fraud.