Lexis Intelligent Assistant Recommendations Deep Dive

Abstract

A legal matter is any inquiry or dispute regarding the rights or obligations of a party in relation to an agreement. Legal matter management refers to a set of activities that need to be completed by In-House Counsels (IHC) within a company’s legal department to resolve such inquiries. The main components of a legal matter resolution process include information research, document management, and team collaboration. Existing templates and documents from past related matters are a key part of this process. IHCs spend a considerable amount of time finding templates and relevant documents for their active matters. In this work, we present a search-based Document Recommendation (DR) approach to assist the IHCs.

Our approach automatically carves out inputs from an IHC’s matter workspace for DR engine. For existing and new matters, matter metadata (such as title, description, and matter type) and information from attached documents (such as document name, authors, and summaries) are examples of extracted inputs for DR. After the automated extraction, the input goes through a pre-processing step where we use key phrase extraction and entity recognition to build feature representations, then we vectorize these features. Next, we use several combinations of lexical and semantic search to recommend the most relevant documents and templates to the IHC for the given active matter. Finally, we present the recommended documents in the IHC’s workspace for easy browsing or inclusion to the current matter.

Keywords: Document recommendation, legal matters, lexis intelligent assistant, cognitive search