LecNet- A Legal Citation Network Benchmark Dataset

Legal document analysis plays a critical role in modern judicial systems, particularly for case retrieval, classification, and recommendation tasks. While datasets like COLIEE have significantly advanced the research in this domain, the lack of large-scale datasets tailored to the Indian judicial system has limited progress. To address this gap, we introduce the Indian Legal Case Citation Network (LeCNet) Benchmark dataset as the first of its kind for the Indian judiciary. The dataset is a gold standard and expert-curated in itself as it has been created using the mentioned citation cases from a source document. LeCNet comprises 26,308 nodes representing case judgments and 67,108 edges representing citation relationships between the case nodes. Each node is described with rich features of document embeddings that incorporate contextual information from the case documents. In this paper, we consider link prediction for a legal citation recommendation. We have performed extensive experiments on the dataset with different AI models for dataset validation. The Mean Reciprocal Rank (MRR) metric is used for model evaluation. Our experiments show promising results for graph structure models that are capable of inferencing dynamic relationships between nodes effectively. The obtained results also demonstrate the utility of our dataset highlighting the importance of graphical networks over textual representations.