- Legal Futures - https://www.legalfutures.co.uk -

Using AI to sort through foreign language documents

By Legal Futures’ Associate Knovos [1]

The ability to quickly understand and then pare down a universe of collected data for review is an essential component of many investigations. This becomes even more complex when documents in need of review contain character-based languages. To avoid the high cost and amount of time associated with human translation and document review, international law firms are now employing artificial intelligence (AI) to simplify the process.

Knovos

The benefits of ECA

Analytic tools used during early case assessment (ECA) [2] help legal teams significantly reduce data volume and conduct exploratory analysis in the early stages of eDiscovery. This empowers litigation professionals to implement the most cost-efficient strategies, increasing productivity while maintaining quality and consistency.

ECA technologies can provide quick access to critical information about case data, with filtering and search capabilities that shrink the size of the data set before sending it to review.

No longer lost in translation

When using an ECA tool on multilingual data, both the tool and its underlying analytics engine must be capable of performing functions such as near-duplicate identification and concept searching on data sets with multiple languages.

Near-duplicate detection ensures that all possible versions of a single document can be quickly identified and clustered together for consistent treatment. In addition, the analytics engine powering the solution’s predictive coding functionality must be capable of performing advanced searching and multidimensional analytics on foreign language documents.

AI and advanced analytics help significantly expedite the processing and review of litigation data, while eliminating the potential for human error.

Korean case study

International law firm Hogan Lovells reached out to Knovos [3] for help with a South Korean client involved in a U.S. Department of Justice investigation. The firm had collected more than 10 million documents, including both Korean and English language records. To meet their client’s deadline and budget, Hogan Lovells needed a solution that could drastically minimize the amount of data reviewed.

By using eZReview, Knovos’ end-to-end eDiscovery [4] platform, the firm was able to automatically generate translations of the Korean documents for quality control and review by native Korean speakers. The team also performed text searching on the Korean documents to determine which were in need of review.

The speed and efficiency of eZReview resulted in a significant reduction of documents, with only 85,000 of the initial 10 million records ultimately selected for attorney review. In the end, Hogan Lovells estimates that eZReview helped save more than 8,300 hours and $400,000 in review expenditures.

Summary

With document review being the largest expense of eDiscovery, reducing the amount of data reviewed is the single most successful strategy toward cost reduction. In addition, when armed with a thorough understanding of the makeup of data, information-driven decisions can be made pre-review about how the eDiscovery process will be handled.

Litigation professionals require AI-powered ECA tools that not only address these needs in a fast, intuitive manner but are able to apply analytics and predictive coding in cases involving foreign language documents.