Summary of Towards Explainability and Fairness in Swiss Judgement Prediction: Benchmarking on a Multilingual Dataset, by Santosh T.y.s.s et al.
Towards Explainability and Fairness in Swiss Judgement Prediction: Benchmarking on a Multilingual Dataset
by Santosh T.Y.S.S, Nina Baumgartner, Matthias Stürmer, Matthias Grabmair, Joel Niklaus
First submitted to arxiv on: 26 Feb 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This study investigates the explainability and fairness of Legal Judgement Prediction (LJP) systems, specifically focusing on the Swiss Judgement Prediction (SJP) dataset. The authors curate a comprehensive collection of rationales from legal experts for 108 cases in German, French, and Italian. They employ an occlusion-based approach to evaluate the explainability performance of BERT-based LJP models, including monolingual and multilingual models, as well as those developed with data augmentation and cross-lingual transfer techniques. The study finds that improved prediction performance does not necessarily lead to enhanced explainability performance, highlighting the importance of evaluating models from an explainability perspective. Additionally, the authors introduce a novel evaluation framework, Lower Court Insertion (LCI), which quantifies the influence of lower court information on model predictions and exposes current models’ biases. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This study looks at how well computer systems can understand and make sense of legal judgments. It’s important to have these systems be trustworthy and transparent because they’re based on factors that might not be relevant or could be sensitive. The researchers used a special dataset called Swiss Judgement Prediction (SJP) to test how well different models could explain their decisions. They found that just because a model gets the answer right doesn’t mean it’s doing so in a way that makes sense. This matters because we want our computer systems to make fair and reasonable decisions. |
Keywords
* Artificial intelligence * Bert * Data augmentation
