Situation

Sky Discovery assisted a client in a large dispute between companies with global offices. The opposing side disclosed over 50,000 documents to our client during the discovery process.

During a high-level review, it was discovered that many documents in the incoming discovery were in languages other than English. The legal team needed to identify and translate non-English documents so the legal team could undertake a meaningful and efficient review.

Solution

Sky Discovery first deployed a specialist language identification tool available within our review platform that identified, without any manual input, over 13,000 documents that contained foreign languages. Once identified, we were able to offer up some solutions around translation.

Due to cost, time, and security pressures, the legal team made the decision to use an AI-powered automatic translation tool. Sky Discovery extracted foreign language text from each document and sent it to the processing agent for translation. The resulting translated text was then returned and applied to a field for each document.

After translation, the text became searchable and could be subjected to analytics. This allowed the identification of near-duplicates and enabled the application of other AI-based workflows like Continuous Active Learning. These tools and workflows helped to accelerate the review of documents that were considered important.

Although the translation process was not as accurate as that completed by a human translator, it did not have to be entirely precise but rather completed quickly. A manual translation provided by a human was undertaken for documents deemed highly important.

Impact

Our work in designing this translation workflow allowed our client to:

• Quickly and accurately identify foreign language documents with little to no human input, saving time and money

• Undertake a translation process quickly and efficiently and to a level of accuracy for the legal teams to begin a meaningful review sooner

• Use the translation outputs to leverage other advanced tools that help to remove documents, highlight similar documents and identify documents that were a priority

• Push any very important documents toward a more accurate translation workflow, which allowed them to get the best understanding of the document content without the time and expense of applying this workflow to all documents in the set.