Inteligência Artificial para a detecção e classificação de mensagens de texto relevantes em evidências criminais
In a criminal investigation, it is common to gather many devices such as smartphones, laptops, computers, and other devices belonging to suspects. The forensic analyst, the professional responsible for analyzing these evidences, has searched for clues that incriminate or innocent the accused. Many o...
Na minha lista:
Autor principal: | |
---|---|
Outros Autores: | |
Formato: | bachelorThesis |
Idioma: | pt_BR |
Publicado em: |
Universidade Federal do Rio Grande do Norte
|
Assuntos: | |
Endereço do item: | https://repositorio.ufrn.br/handle/123456789/45232 |
Tags: |
Adicionar Tag
Sem tags, seja o primeiro a adicionar uma tag!
|
Resumo: | In a criminal investigation, it is common to gather many devices such as smartphones, laptops, computers, and other devices belonging to suspects. The forensic analyst, the professional responsible for analyzing these evidences, has searched for clues that incriminate or innocent the accused. Many of these clues can be found in texts. However, finding these texts is a tired and slow process because a unique smartphone can contain thousands of messages, and in an individual investigation can contain tens or hundreds of smartphones. In order to minimize the effort to find the suspect and relevant text messages and consequently speed up and improve the forensic analyst job, this work proposes an Artificial Intelligence to detect and classify relevant text messages in the context of criminal investigations. Therefore, some Machine Learning algorithms are investigated to accomplish this task, such as Random Forest, Logistic Regression, Support Vector Machine (SVM), and XGBoost. Due to the context of these messages, this work is supported by Ministério Público do Estado do Rio Grande do Norte (MPRN), which provides all needed data, tools, and computational resources for the development of this work. The institution helps build the database to represent the set of relevant text messages and define the classes of messages represented by the set of relevant messages. A WEB System was developed to execute in the MPRN local environment to the analysts and other professionals mark and classify the data to the dataset. |
---|