21 Feb 2020

FLOW: Forensics, technology and the future of legal services 10: Technology assisted learning

15,000 documents for review as part of discovery might sound a lot, but not when you started with over 17 million. Daniel Heywood looks at how technology-assisted review can work to tame the beast of document review.

Other FLOW videos

Related Knowledge

Get in Touch

Get in touch information is loading


Clayton Utz communications are intended to provide commentary and general information. They should not be relied upon as legal advice. Formal legal advice should be sought in particular transactions or on matters of interest arising from this communication. Persons listed may not be admitted in all States and Territories.


Our clients are continuously faced with issues with large data sets, now this can be with issues such as regulation or also responding to litigation in e-discovery.  Now one issue that we do have is how do we get through that data in an efficient and effective manner when it is large and complex and we don't have much time to do it.  The preferred approach that we have is to use technology.  One recent example we had with a client was they needed to go through a huge amount of data and put it into 12 different categories.  In that example we were sourcing our information from email, shared drives, local drives and also from customer relationship management systems. 

Using the various technologies that we have we started with 17 million documents, we got that down to 135k through keyword searches and then from there we were able to use technology assisted review to further bring that down to only 15,000 documents which we had to manually review. 

The use of technology assisted review in this case help reduce error from humans reviewing large sets of data and then it created a paper trail which we could go back to and then look at what had been happening throughout the whole process.  Technology assisted review is designed to pull conceptually related documents together and prioritise those documents for review, it targets a small set of documents that is conceptually relevant and uses documents that are not relevant that we can potentially disregard to the left side of the chart. 

The chart on the bottom tells us when a relevant rate starts to drop and when we can potentially stop our review and look at the discarded sets and make a decision whether we want to continue or not.  Information obtained from a set of documents become our scope of searches for additional data sources and details relating to a single customer can then be linked.  For example in a document we can identify a customer and then in other documents where that customer exists we can link those documents up to a group and have them formed together. 

What we are able to do as well is identify the valid voice recordings made by those customers to the company and pull those together in the one review stream so through the use of technology we increased our accuracy and thoroughness of the review.