Wolters Kluwer Deutschland Semantic Search

A highly advanced tool to simplify the retrieval of complex legal information using a combination of sophisticated technologies innovatively.

Industry Challenge

Wolters Kluwer Deutschland GmbH (WKD) provides professionals in the legal/tax domains access to actionable and intelligent tools, empowering organizations to make better decisions.

However, the legal professional clients of WKD were finding it labour intensive to search for quality documentation relevant to a given case using the current tools. Hence, the primary goal of the collaboration with ADAPT was to solve this problem to better improve efficiency and productivity.

The project required a thorough study of the precise needs of legal professionals for whom the system was to be designed. This was then applied to develop a carefully planned evaluation framework using individual scientific components and their use in combination.

The ADAPT Solution

Since this was a highly complex and previously unproven project, the goal was to first develop a proof-of-concept for use by WKD’s legal professional clients. With the aim of improving efficiency of the search tools, a technical investigation was carried out on the effectiveness of state-of-the-art information retrieval methods and their novel combination with WKD’s domain-specific knowledge structure.

After careful analysis of the needs and working practices of WKD’s legal professional clients, a workflow for the prototype system was developed. This was designed to take informal case notes by legal professionals as input search queries and identify particular legal concepts within these notes. These were then used as input for the search algorithms under investigation to output ranked lists of documents potentially useful to legal professionals.

The project was highly interdisciplinary and required inputs from a number of researchers with diverse expertise. Since some aspects of progress were constrained by the availability of the expert researcher, the project was spanned across a lengthy duration. Nonetheless, it was a valuable scientific project to be part of, for all team members even though it had equally unavoidable output risks.

Key Outcome

Research in improving efficiency of search using text contents of documents, has indicated that efficiency is achieved by document ranking based on a document fragment index rather than search of full documents.

Key Learnings

The complexities of professional search go beyond those present in more public-facing search problems which form the basis of most academic research in information retrieval. It would be interesting to further investigate whether the strategies used in academic research can be extended for use in the area of professional search.