A Framework of Multi-Stage Classifier for Identifying Criminal Law Sentences

A Two-Stage Classifier That Identifies Charge and Punishment under Criminal Law of Civil Law System

“…The benefit of this is at least two fold: a reduction in the training data constraint and conformance to the legal decision process in the real world. [2] The first stage is to identify the case-level attributes based on obtained fact-level attributes. This is very important and requires computing schemes that can comprehend the subtleties way of human judgment.…”

“…The information gain is based on the entropy function, as shown in Eq. (2). The function f req(C j , T ) calculates the frequency that case C is classified as T j .…”

“…By the legal process of the Civil Law system, criminal facts can be collected, investigated, and collated into a more abstract layer that conforms to the elements and frameworks in the law articles. To reflect this nature, Thammaboosadee et al [2] proposed sequential data analysis for the judicial process as follows: (i) criminal facts are collected and formalized into fact-level data, (ii) fact-level data are classified into diagnostic issues (case-level data), and, lastly, (iii) case-level data are analyzed and classified into legal elements (legal-level data), for correspondingly applicable articles for identifying charges and the range of possible sentences. This systematic inferencing process yields decisions that can be defended by the system [3].…”

Assessment of text clustering approaches for legal documents

“…Um dos resultados desse trabalho indica que os fatos são o fator preditivo mais importante na realização dessa tarefa. [Chen et al 2019], [Kowsrihawat et al 2018] e [Thammaboosadee et al 2012] também utilizaram os fatos dos casos para predizer resultados das sentenças dos julgamentos através de diferentes modelos baseados em Deep Learning, Redes Neurais e Árvores de Decisão.…”

An empirical analysis of Brazilian courts law documents using learning techniques

“…The application of intelligent techniques for machine learning as facilitator of the decision-making process of magistrates of Brazilian judiciary could increase productivity of Courts [Coelho 2017], considering capacity of processing legislation and documents that would be dedicated to computer [Thammaboosadee et al 2012], and would give greater equal treatment of cases (isonomia in Brazilian Portuguese), since it would replace the human components of overload and absorption [Fonseca et al 2018]. There is a lot of research and projects to deliver intelligent products into the judicial process as [Thammaboosadee et al 2012, Fersini et al 2009, Branting 2017].…”