Cognitive forensic content analysis
Martin Geyer engaged IBM Premier Business Partner BIConcepts IT Consulting GmbH to deploy a cognitive content analysis and investigation tool. The solution ingests data from multiple formats into a corpus of knowledge. The fact that the solution enables users to define their own categories allows the model to closely reflect the parameters of the case. Natural language processing (NLP) algorithms conduct semantic searches for text patterns. The core function of the solution is to identify instances within documents that correspond to user-defined parameters. In the case of a price-fixing search, the solution can find all documents related to the topic of prices even if only numbers, currency codes or expressions, such as “best price guarantee,” appear. Moreover, a discount agreement—a red flag in such a case—will be discovered even if the word “discount” is not explicitly mentioned, but percentages and rebates are discussed. By identifying and flagging high-relevancy documents, the solution guides investigative specialists to where the evidence is to help them more quickly and efficiently build the case.
Efficiency significantly improved
Martin Geyer significantly improved the efficiency and accuracy with which it can review complex legal documents. The company reduced its review process by up to 50 percent compared to its previous manual process, along with commensurate reductions in the costs associated with these searches. The solution also reduced the delays that typify complex legal proceedings, which often hinge on finding information embedded in huge volumes of documentation. In addition to faster resolution of legal cases, the solution makes it easy to apply cognitive insights to future cases, further strengthening the firm’s business case.
The solution is game-changing for the company because it fundamentally alters its strategic core competency: that is, its ability to piece together complicated fragments of evidence, gleaned from huge volumes of unstructured data, to support legal proceedings. Large-scale cognitive forensic analysis instantly shifts the company’s resource focuses toward the more sophisticated analyses of relevant documents, conducted by experts.
Previously, the company relied on as many as 20 assistants to read all the documents and flag them as relevant or irrelevant. Only when relevant documents were flagged and set aside did the experts step in to conduct more detailed analysis. The new system automates and dramatically compresses the initial forensic analysis cycle.
The solution represents a system of cognitive discovery because it applies NLP algorithms to understand the semantic context of a wide range of unstructured, text-based legal documentation and correspondence.