KordaMentha KordaMentha is an Australian advisory and investment firm that provides specialist forensic, real estate, turnaround / restructuring and investment management services. The business was formed in April 2002 by Mark Korda and Mark Mentha. Located at
Level 24, 333 Collins Street, Victoria, Australia.
Phone: +61 3 8623 3333

Forensic blog


Australia Joins the TAR Trend

by Craig Macaulay | Dec 06, 2016

Our earlier publications have followed the use of Technology Assisted Review (TAR) – mostly overseas[1].

 

But now there’s significant action in Australia.

 

Firstly in Money Max Pty Ltd v QBE Insurance Group Ltd VID 513/2015 the Federal Court has ordered the Respondent to report to the Applicant an explanation of its use of TAR in providing discovery.

 

More significant is McConnell Dowell Constructors v Santam Ltd Ors 2016 VSC 734. Justice Vickery said:

 

The central issue for the Court is this: given the enormous number of documents generated by the proceeding, how discovery should be managed, consistently with the principles of proportionality and Section 9 of the Civil Procedure Act 2010 (Vic.) (the ‘CPA’) which requires the Court to further the Overarching Purpose of the CPA when making orders or giving directions. Section 7 of the CPA provides that:

 

(1) The overarching purpose of this Act and the rules of court in relation to civil proceedings is to facilitate the just, efficient, timely and cost effective resolution of the real issues in dispute.

 

…Standard Operating Procedure (TEC[2] SOP 5 [TAR]) was issued as an interim measure to provide guidance to parties and legal representatives for the review of documents by using TAR (Technology Assisted Review or Predictive Coding). This development was in response to the unique challenges presented to the TEC List in the management of construction and engineering cases of the scale of the present. Large document production is particularly prevalent in the field of construction and engineering projects, where technology is used to generate and disseminate project information to a large number of participants. With the commonplace use of ever more powerful computers in large construction and engineering projects, collection and storage of electronic data is growing at an exponential rate. The end result is that an enormous volume of information is now created, exchanged and stored electronically, as ‘electronically stored information’ or ESI. This phenomenon calls for special management in the event of litigation or arbitration in a construction or engineering case.

 

The orders allow for McConnell to review its 1.4 million documents using TAR. This is in line with some overseas judgements. But Justice Vickery has gone further by appointing a special referee to manage the process.

 

The special referee has ordered the following:

 

1. Use of predictive coding:

      a. The matter will be using FTI’s Ringtail software to perform predictive coding;

2. Creation of initial training and comparison set:

      a. A random sample of 1,535 documents will be created as the comparison or control set to be reviewed and tagged           by the Plaintiff’s and Defendants’ legal teams.

      b. An initial training set (seed set) of 1,335 documents will be created as follows:

             i. 1,000 documents randomly selected by FTI,

             ii. 200 relevant documents identified by the Defendants, and

             iii. 135 irrelevant documents identified by the Defendants;

3. Training to improve the model:

      a. Continue to create training sets using 1,000 to 1,500 documents until the model is stabilised and the Plaintiff and            Defendants are comfortable,

       b. Both parties expect that it will take 10 to 15 rounds of training to achieve model stabilisation,

      c. The Plaintiff will review the documents and provide feedback in a spreadsheet with document ID, document title            and whether the document is relevant or irrelevant, and

4. Creation of the validation set:

      a. Create a sample based on 95% confidence level with +/- 5% margin of error. The exact size of the sample is a            function of the prevalence of responsive documents throughout the corpus, and

      b. The Plaintiff will review the documents and provide feedback in the form of the final recall and precision document          counts;

5. Methods and indicators for assessing the performance of the predictive coding:

      a. The training is to continue until the desired level of recall and precision are achieved, and

      b. The desired level of recall is 80%. The level of precision is not detailed in the orders;

6. Identification of the final review population:

      a. The IT services firms, representing the Plaintiff and Defendants, will monitor the process until the predictive coding          model is stabilised. Then, subject to quality control checks, the responsive documents and families will be identified          as the discovery population.

 

The judgement has a useful glossary of terms.

 

It will be very interesting to see how this case will affect the use of TAR in Australian Courts: the lack of any Australian case law has made many lawyers nervous about using TAR, so perhaps we are about to see the floodgates open?  Watch this space!



[2] TEC – Technology Engineering and Construction List of the Victorian Supreme Court

 

KM Private Clients  |  Staff Access  |  Privacy Policy  |  Legal Status  ©  Copyright . KordaMentha.