DoSSIER@COLIEE 2021: Leveraging dense retrieval and summarization-based re-ranking for case law retrieval

In this paper, we present our approaches for the case law retrieval and the\nlegal case entailment task in the Competition on Legal Information\nExtraction/Entailment (COLIEE) 2021. As first stage retrieval methods combined\nwith neural re-ranking methods using contextualized language models like BERT\nachieved great performance improvements for information retrieval in the web\nand news domain, we evaluate these methods for the legal domain. A distinct\ncharacteristic of legal case retrieval is that the query case and case\ndescription in the corpus tend to be long documents and therefore exceed the\ninput length of BERT. We address this challenge by combining lexical and dense\nretrieval methods on the paragraph-level of the cases for the first stage\nretrieval. Here we demonstrate that the retrieval on the paragraph-level\noutperforms the retrieval on the document-level. Furthermore the experiments\nsuggest that dense retrieval methods outperform lexical retrieval. For\nre-ranking we address the problem of long documents by summarizing the cases\nand fine-tuning a BERT-based re-ranker with the summaries. Overall, our best\nresults were obtained with a combination of BM25 and dense passage retrieval\nusing domain-specific embeddings.\n