Features - Business

Challenges of e-disclosure in construction cases

As the use of technology has changed the way we communicate, and the volume of documents created every day has drastically increased.

It is estimated that 156 million emails were sent every minute last year. Add to that the new methods of communication developing and being favoured by companies and individuals, and it is easy to understand why the data created by companies during the course of a project is constantly growing – but as a result so is the task of e-disclosure. Much has been done in recent years to improve the efficiency and usefulness of the disclosure process in arbitration and court litigation.

Maren Strandevold, Associate at Haynes and Boone CDG,

Maren Strandevold, Associate at Haynes and Boone CDG,

This article, from Maren Strandevold, Associate at Haynes and Boone CDG, will discuss the usefulness of the tools and techniques available to parties in a construction dispute to deal with the data explosion.

Disputes in the construction sector are typically document heavy, particularly where the dispute includes a claim for delay on a big project. The document count can be in the hundreds of thousands and that in itself presents a number of challenges. The nature of the work creates further challenges: disputes are usually technical in nature and the relevant evidence is likely to include spreadsheets, drawings and photographs. Managing such documents, reviewing them for relevance and providing disclosure can be a real challenge and lawyers engaged in this type of work have to look at the tools available and consider carefully the most appropriate approach. Making the wrong choice can make the exercise both more costly and far more time consuming.

The basic toolkit

Whilst there are still lawyers who would prefer to print out all the documents and hand over hard copies, it is widely accepted that disclosure today requires the use of an e-disclosure platform that can process documents much more efficiently. A processing tool will also typically be used prior to uploading documents to a platform. In their most basic form, this allows you to:

  • Sort documents by file type
  • Run basic searching based on keywords and metadata
  • De-duplicate documents.

Using these simple tools, you can start assessing the data available early on. It may be sensible to hive off certain types of documents, for example drawings, for review by expert witnesses. Similarly, it may be useful to ring fence photographs for human review as searching is difficult unless the photographs have been saved with meaningful titles. Even at this early stage, you can start prioritising the review and ensuring that the most appropriate resource is used.

Keyword searching

Keyword searching remains very popular as a method of locating relevant documents and ensuring that a “reasonable search” has been carried out. In the English courts, the forms that must be exchanged prior to the first case management conference (even under the pilot scheme for the Business and Property Courts which is due to start in January 2019) encourage parties to think about keyword searches at an early stage.

Keyword searches can be used to narrow down the document population and reduce the quantity of documents that have to be reviewed during the disclosure process but it is essential that you know your document population and what you might be looking for, as otherwise this method has its drawbacks. It is a very blunt tool and unless you understand the terms and phrases typically used on that particular project, you may end up with a significant volume of documents to review, a significant proportion of which may well not be relevant. Misspellings must also be accounted for, otherwise key documents will not be caught by a search, which can lead to important information being missed.

Analytical tools

The main alternative to keyword searching followed by manual review of documents that is often proposed is full-on predictive coding, which is discussed below, but this fails to consider the variety of other options available. For example, if you have a huge volume of emails, but know that a significant portion are irrelevant, it usually helps to see the domain listing – that is a list of all the email domains in the population. This information can be used to exclude irrelevant sources, such as news subscriptions, from the pool of data that requires review.

Alternatively, you could identify the key domains involved in the project and focus on reviewing correspondence passing between those domains, ie: you focus your review on emails passing between @employer.com and @contractor.com.

There are also some more sophisticated tools such as clustering and concept tools. These can be particularly useful if the claim comprises a number of different issues, for example if you have a delay claim with multiple relevant delaying events. The clustering and concept searching works to find patterns in the text contained in documents. Concept searching allows for more dynamic searching than keywords while clustering may assist by grouping together documents with similar content, so that documents relating to specific issues may be identified and reviewed together.

Predictive coding

There has been a lot of hype around predictive coding, with it being hailed as a cost-effective yet accurate way of categorising large numbers of documents. In construction disputes, however, the feeling has typically been that it is not appropriate. Predictive coding is good at identifying single issues and less good at dealing with technical documents (spreadsheets and drawings) and multiple issue cases, such as a delay case with hundreds of separate delaying events arising out of entirely separate factual matters.

But is it true that this renders predictive coding useless in a complicated delay claim?  I think not. As with a lot of the other technology available, it is very much a question of how you use it. For example, you can use the technology to rank documents so that you can focus the human review primarily on the most highly ranked documents. In addition, you would ideally conduct a review of a sample of the lower ranked documents, to find out if these need to be reviewed further or can all be considered irrelevant.

The existing technology can therefore be of use, even in cases with multiple issues and large volumes of technical documents. However, the technology is constantly evolving. Some e-disclosure providers have seen the need for technology to cope better with non-textual documents and are in the process of developing technology that can carry out predictive coding of images. There is already technology for facial recognition and image searching and this is now being harnessed for use in e-disclosure. This is largely untested at present, but could be the first step towards better management of technical documents.

The use of predictive coding on the right cases and in the right way can greatly reduce the amount of manual review required and the associated time costs. Computer-aided review is often more consistent than human review and it has the potential of reducing the overall cost of the exercise. However, this can only be achieved if those using the tools understand how to use them correctly and consider their application on a case by case basis.

If you would like to read more articles like this then please click here.