A.I. driven review platforms seem to be a hot trend. First Reveal Data, followed by Relativity’s acquisition of Text IQ and now Exterro releasing a new review offering based on ‘deep learning’. I always want to understand the tech and mythologies underlying solutions. Ajith Samuel and Bill Piwonka at Exterro were kind enough to walk me through Exterro’s investments in A.I. going back to their 2018 Smart Labeling offering. Before we dive into the techie deep end, I asked for the highlights on the new review platform.
- 300% overall performance improvement in time-to-review processing.
- eDJ-While I would love to think that we are past the Processing Performance Wars a decade past, ballooning custodial collections and slow M365 Advanced eDiscovery indexing can delay investigation and data breach responses by weeks. Time-to-review is a key SLA metric to corporate Legal-Compliance-Security under pressure in post incident response scenarios.
- New review interface focused on usability, click reduction and intuitively available features to ease adoption and maximize efficiency.
- eDJ-Usability is key to overcoming prospects hesitation to switch technologies. If Exterro wants to take market share from Relativity their offering has to be intuitive and easy to learn. See before/after images below.
- Maturity features that support large, complex matters managed by firms and contract review providers.
- eDJ-Exterro’s initial review offering was focused on their corporate customer review scenarios. Exterro Review is meant to open the door for firms and providers looking for a Relativity alternative to tackle very large discovery challenges. See before/after images below.
- Neural machine translation, a new feature that enables users to search for, identify and translate foreign language documents swiftly and securely without interrupting the review workflow. Smart labeling and reviewer suggestions that call attention to documents that a reviewer may have miscategorized as responsive or not; and more sophisticated entity recognition that can identify, for example, name misspellings.
- eDJ-The value proposition on these are pretty self-explanatory.
Now for a peek behind the black box ‘deep learning’ curtains. Many of us tend to view analytic systems as discrete tools. It is easier to explain individual text mining/clustering methods such as Latent Semantic Indexing, sentiment analysis, predictive phrase completion, Natural Language Processing, etc. than the complex application of multiple methods found in Continuous Active Learning. Sometimes the individual proprietary or open source underlying systems are less important than how they are applied to data sets and review decisions.
Exterro began building their A.I. infrastructure and customer focused neural language models in 2014-2015. They wanted to support a wider range of usage scenarios from GRC through to eDiscovery. As AJ pointed out, generic language models may support your Google predictive search terms, but they must be adapted to incorporate matter concepts, contextual aspects, entity facets and other key elements of investigations, breach remediations and discovery topic relevance.
My interpretation of Exterro’s deep learning is that they have gone for a more holistic approach to A.I. supporting diverse customer usage scenarios. The combination of complex entity extraction, data mapping for source context, pattern/anomaly detection and concepts in context may indeed give customers a more practical ability to generate narrative storylines. Time and market share will show if their long term strategy has paid off. I am just happy to have more realistic options for client RFPs.
Greg Buckles wants your feedback, questions or project inquiries at Greg@eDJGroupInc.com. Contact him directly for a free 15 minute ‘Good Karma’ call. He solves problems and creates eDiscovery solutions for enterprise and law firm clients.
Greg’s blog perspectives are personal opinions and should not be interpreted as a professional judgment or advice. Greg is no longer a journalist and all perspectives are based on best public information. Blog content is neither approved nor reviewed by any providers prior to being published. Do you want to share your own perspective? Greg is looking for practical, professional informative perspectives free of marketing fluff, hidden agendas or personal/product bias. Outside blogs will clearly indicate the author, company and any relevant affiliations.
Prior Review management screen