Improving the performance of Pakistan’s judicial system and its ranking using data analytics

Abstract

A large number of lawsuits are pending with various courts in Pakistan. This large num ber of pending lawsuits indicates the ineffectiveness of the justice system. An ineffective justice system contributes to decreasing the rank of the country in the World Justice Project (WJP), which is an international body, working for the advancement of the rule of law worldwide. Unfortunately, the rank of Pakistan in the WJP remained constantly low since Pakistan became a member of the WJP. This thesis aims to improve the effec tiveness of the justice system in Pakistan using data analytics. The proposed solution contains two parts. The first part is a legal dashboard that provides insights into vari ous lawsuits registered with the Supreme Court of Pakistan and its sub-courts. On the basis of these insights, higher management of the judiciary will be able to monitor their performance and devise strategies to improve weak areas, which can help improve the rank of Pakistan in the WJP. The subjective evaluation of the effectiveness of this part was done by legal practitioners and students of Islamabad-based renowned law schools. The majority of the evaluators were of the opinion that the proposed solution will be helpful in improving the effectiveness of the country’s justice system. Past judgments act as a reference in the current court proceedings. That’s why legal practitioners spend a significant amount of time reading past judgments. The majority of the time, after reading a judgment, it is found that the judgment did not contain the information of interest. This results in the wastage of time and effort. Topic modeling is an important technique for getting insight into textual documents. The second part of the solution is a topic modeling technique that will help legal practitioners find whether judgment(s) is of their interest or not. We implemented three topic modeling techniques i.e., LDA, LSA, and HDP on legal judgments. The performance of the topic modeling techniques was evaluated on the basis of the coherence CV score. The HDP model outperformed all other techniques for extracting topics from legal judgments.