Information extraction from case files and regulations

A unified knowledge base creation by extraction of data from both regulations and case files.

Ujwal Narayan, Pulkit Parikh, Kamal Karlapalem, Vrinda Lakhotia, Sandeep Dash

Case files and adjudication orders have a lot of information that are not present in the regulations. These are often used as precedents when a similar situation arises, and thus for any downstream task such as penatly prediction it becomes important to understand and extract the information present in the case files. Similarly, regulations are also explored through the usage of case files which lets you know the scope of the regulations and how and where it is applied. Thus we have two different knowledge bases: regulations and case files, and it becomes necessary for systems to leverage this information. This is a step towards that direction. We propose a graph based architecture that unifies the regulations and case files dealing with the same concepts.

Data: Here

Slides: March 9th Slides