A Machine Learning Based Prediction System of Medical Laws Judgment Using Statute-Classified Decision Tree with Text Similarity

• Main Authors: CHI,HSING-CHEN, 紀幸辰
• Other Authors: CHEN,YUH-SHYAN
• Format: Others
• Language: en_US
• Published: 2018
• Online Access: http://ndltd.ncl.edu.tw/handle/gxyh35

碩士 === 國立臺北大學 === 資訊工程學系 === 107 === The retrieval of legal documents takes a considerable time, in which predictive accuracy of the file is an important issue in the law domain. The legal documents (judgments) record the legal reasons, processes and applicable statutes of the law. Because the verdict did not clearly indicate the outcome of the lawsuit. The result need to analyze the law and corpus through the comparison between the judgments. In this thesis, we propose the prediction system of medical laws based on machine learning. The system predicts the result of lawsuit through the statute and the corpus that the result is the plaintiff victory or the defendant victory. The application area of this system is assist with the lawyers. The system saves the lawyers' text retrieval time and providing the high-accuracy predictions. The first priority of a predictive system is to analyze the statute of the judgment. When the statute cannot make an accurate prediction, the system considers the relationship between the texts, and use the method of text similarity. The system judges the outcome of the lawsuit according to the case with high similarity. To illustrate the performance achievement, the mathematical analysis and the simulation results are examined in terms of the accuracy, the precision, the recall, and the MAE.