ISM@FIRE-2013 Information Access in The Legal Domain
Ambedkar Kanapala, Sukomal Pal · 2013
This paper describes the work that we did as a participant in the task of Adhoc retrieval from legal documents (Consumer Law and Hindu Marriage & Divorce Law) at FIRE 2013. We managed two runs using the Indri search retrieval system as official submission. However, post-submission, we experimented with ten different models using the TERRIER system. With the available qrels, we could evaluate only the runs pertaining to Consumer Law. Initial results have shown substantial improvement in retrieval performance. Our experimentation has also resulted in preparing a cookbook for optimal parameter-setting for a number of Terrier models. In Identification and Classification of Propositions in Court Judgment (parse each judgment into individual propositions), we have written a program which takes a single file as an input from the corpus and breaks the data of the given file into individual sentences. In summary, ours was a rich learning experience which needs to be continued to gain further insights. Also, we believe as pilot track, information access from legal domain is promising enough to be continued further as it has potential of being a fertile research area, specifically in Indian context.