Concept Search for Legal Review

Concept searching goes beyond synonym or keyword matching to include all documents that describe the same subject matter or concepts, regardless of the specific terms or words used.

For example, concept search understands that the words “terminating” and “firing” and also the phrase “ending association with” are all describing the same idea or meaning. The ability to harness the power of true concept searching and advanced analytics is critical to the ability to perform an intelligent, thorough and cost effective review.

How it works:

Evident’s latent semantic indexing derives meaning from documents through a rigorous mathematical analysis of the relationship between terms across all of the case documents. It requires no outside assistance from people or dictionaries, and this enables concept searching to keep up as language evolves.

When to use Conceptual Search:

Email culling and keyword searches are good for finding domain names, date ranges organization names, and documents on small cases. Cases that have millions of documents to review require a multi-dimensional search capability like Conceptual Search, with Dynamic Clustering and Categorization to identify relevant documents. Conceptual Search allows the reviewer to determine themes within the data, which is vital to finding critical responsive documents.

Concept Search with advanced analytic engines can be used as part of a pre-review analysis to organize related documents into concept-based folders. This enables the legal team to identify what issues or concepts are present in the overall collection, and your top reviewers can then focus on the key folders. This filtering capability reduces review time, improves review accuracy, and provides significant savings on overall litigation costs.

The Defensibility of Evident Discovery Software Solution:

Evident Discovery’s Conceptual Search engine uses a patented and proven technology called Latent Semantic Indexing. Its patented software routines yield a mathematical representation of documents that is inherently more accurate than human review.

When conducting any search, documents being returned (or not) will fall into one of three categories: relevant, non-relevant, and documents missed (not returned but relevant). The measurement of accuracy that any solution provides is the combination of Precision and Recall:

• Precision:  The precision rate is the percentage of identified documents that were relevant. For example 100 documents were identified, but only 80 of them were relevant. So precision is 80%.
• Recall:  The recall rate is the proportion of relevant documents identified, compared to the total number of relevant documents that exist. For example the total number of relevant documents is 100, but only 75 were identified. Your recall is 75%.

Evident Discovery concept search has demonstrated Precision and Recall rates of 89% and 88.3% respectively – a magnitude of performance superior to human review.