Neural Legal Judgment Prediction in English

Abstract

Legal judgment prediction is the task of automatically predicting the outcome of a court case, given a text describing the case's facts. Previous work on using neural models for this task has focused on Chinese; only featurebased models (e.g., using bags of words and topics) have been considered in English. We release a new English legal judgment prediction dataset, containing cases from the European Court of Human Rights. We evaluate a broad variety of neural models on the new dataset, establishing strong baselines that surpass previous feature-based models in three tasks: (1) binary violation classification; (2) multi-label classification; (3) case importance prediction. We also explore if models are biased towards demographic information via data anonymization. As a side-product, we propose a hierarchical version of BERT, which bypasses BERT's length limitation. 1 The dataset is submitted at https://archive. org/details/ECHR-ACL2019.

Extracting Outcomes from Appellate Decisions in US State Courts

“…Firstly, the CASELAW4 dataset could be used in a number of prediction models, from more complex LSTMs to transformers to pre-trained language models. Since the problem of legal outcome prediction is a highly complex problem that relies on numerous factors, sophisticated deep learning models show promising results [12,19,25]. Secondly, it is important to further improve outcome extraction, to go beyond the binary system of AFFIRM and REVERSE labels and to move to more granular MIXED cases.…”

“…Previous works focused mostly on European and Chinese law. They include predicting outcomes in the French Supreme Court [18], in the European Court of Justice [14,19], and in the European Court of Human Rights [13,12,15,16], as well as predicting outcomes of criminal cases from the Supreme People's Court of China [10,20,21,22,23,24,25]. However, very limited work focused on the U.S. and U.K. law systems [9,17], and to our knowledge, no attempt has yet been made to predict outcomes for cases from the CAP dataset [8].…”

“…It offers potential to improve access to justice, automate repetitive administrative tasks, reduce legal costs, and bring transparency to judicial procedures [4]. Considerable research effort has recently been devoted to case outcome prediction -the task of predicting the outcome of a court's decision in a particular case (i.e., verdict, sentence, charge, or fine) given the factual background of the case [13,12,17,22,24]. Legal analytics requires a sufficiently large-scale dataset of case information, including facts and outcomes.…”

Judges Are from Mars, Pro Se Litigants Are from Venus: Predicting Decisions from Lay Text

“…• N-gram frequency vectors, for n = 2-4 • Vectors of 250 topic models 5 We compared the performances of six machine-learning algorithms-Naive Bayes, Bayes Net, SMO, JRip, J48, and Random Forest-in 10-fold cross validation. The results of the highest-performing algorithms are shown in Table 1.…”

Data-Driven Prediction of Judgment. Law’s New Mode of Existence?

“…Chalkidis, Androutsopoulos, and Aletras plan 'to expand the scope of this study by exploring the automated analysis of additional resources (e.g., relevant case law, dockets, prior judgments) that could be then utilized in a multi-input fashion to further improve performance and justify system decisions'. 48 Once again, what they call 'a justification' of system decisions is not a justification in the legal sense, but the verification of a causal influence of specified input features on the system's output variable (violation or no violation). The law is, however, not about the causality between specific arguments and their conclusion, which would entail a category mistake.…”