OCTOBER 2013
Eversheds
27
the form of requests for the provision of documents and information.
The US Department of Justice and other US government agencies are still looking at what went on in the Cayman Islands. Further prosecutions or possibly civil actions cannot therefore be ruled out. Predictive Coding – Bringing Humans and Machines Together
So will anyone be using predictive coding technology in Japan, the US, the UK or the Cayman Islands? The technology is not confined to civil or criminal proceedings or indeed to the legal environment at all.
Predictive coding technology is designed to conduct automated searches across a substantial body of electronic information in order to retrieve relevant material. It is more sophisticated than key word searching. Key words define objective criteria with reference to which the software can very quickly search for and retrieve items in the dataset answering those criteria. Predictive coding goes further and enables a range of other criteria to be specified in the automated search. The way this is done is for an experienced lawyer or investigator to work with a sample set of documents drawn from the dataset (or indeed from another source outside the dataset) and to categorise (i.e. code) those documents as relevant or not relevant or as belonging to some other specified category. The categorisations draw on a range of nuanced criteria determined by the experienced reviewer. These human actions might be a grouping of documents as constituting some evidence of a conspiracy or evidence against a particular individual, for example.
The software will then assess the way that the initial sample set has been classified, looking at all the criteria present in those “seeded” documents and automatically search the remainder of the document pool to find similar documents exhibiting the same or very similar criteria. The idea is that the most relevant material can be automatically identified and quickly prioritised to be examined first.
In any investigation, one is looking to build links and identify further lines of enquiry. No doubt reading the initial documents from the whistle blower in this case will have set a number of lines of enquiry running, which would have generated requests for further material from new sources. Once that further material has been produced, it would have been possible to use predictive coding to sort through it quickly with a view to isolating what really matters. Predicting the Prosecution in a Cross-Border Investigation
There is no reason why a prosecutor in the UK could not use this technology in an investigation of this kind, particularly where there are large document sets that have to be reviewed. This is one very valuable way of harnessing the power of computing to save human resources or even just as a method of moving the investigation forward, using an effective combination of man and machine in a strategic manner more quickly. This has the possible knock-on advantage of getting
hold of evidence before the trail goes cold or the evidence disappears.
In a case like Olympus, one might imagine the investigators
studying different types of
electronically stored information, such as corporate email information, very significant volumes of accounting data and historical information going back several years (reports about this case suggest that the payments to the Cayman Islands were part of a scheme to hide losses going back over a long period of time). The metadata (electronically stored information indicating who accessed, created, amended or deleted a document) would be of significant interest too.
The use of predictive coding technology in the US is much more likely, however. The technology has most recently been developed there and has been in use for some time. US investigative agencies are known to have deployed predictive coding software. For example Recommind’s Predictive Coding technology has been used this year by the enforcement division of the U.S. Securities and Exchange Commission. Furthermore, companies in the US traditionally create and store significantly larger volumes of data than might be the case in the UK making technology essential to handle the volumes.
In a cross-border investigation of this kind different data sets will no doubt have been obtained from different sources in different countries and will need to be combined and searched as one global repository. An item by item, so-called “linear”, review would not be an effective approach and would be extraordinarily expensive. Putting predictive coding to use can help any investigator for a legal purpose (be that regulatory, civil or criminal) and enable more focussed enquiries to be formulated and followed early on.
Proactive Fraud Prediction In the Olympus case, corporate entities, rather than any individual, have been pursued. Therefore, the investigation will have to determine what senior management knew about
the suspicious
payments. That in turn will translate into very significant volumes of corporate data. Piecing together an evidential puzzle of who knew what, when, is the essence of a fraud investigation and one that is always difficult to do. Significantly, this case explicitly involved information being hidden from the auditors. Where concealment is involved it is all the more difficult to unravel exactly what happened, when and why, and to then interpret those events as constituting an offence so as to persuade a jury beyond reasonable doubt.
The Serious Fraud Office is not specifically known to be using predictive coding technology. The Financial Conduct Authority has invested heavily in technology to assist its investigations but it is not believed that predictive coding technology has been embraced just yet.
Like all technology predictive coding technology does have some limitations. The Olympus case is an accounting fraud involving concealment in the activities of a company concerning its accounting for specific payments. It may well be that
www.lawyer-monthly.com
accounting documentation will have been isolated and reviewed without the use of predictive coding. However, businesses could use predictive coding to proactively surface potential risks and fraud before it happens. For example, predictive coding can intelligently and proactively identify and recognise clues simply from the tone of an employee’s messages and its context in other communications. Once alerted, the business can then investigate further, either manually if needed, or by taking a “show me more like this” approach – training the system to identify potentially fraudulent material without having to know in advance the precise words used, enabling them to identify fraud before it happens.
controlling the data Explosion Predictive coding technology is likely to not only change the way investigators and prosecutors organise the evidence in criminal cases, but it is likely to transform the way litigation is carried out in the civil world. The challenge facing the legal and commercial environment is the exponential explosion in data volumes. This is a challenge created by technology and one that can only be solved by technology. What will make all the difference is the acquisition of the skills by professionals in harnessing the power of this technology to achieve specified objectives.
The technology is not just for the prosecution. It can work on the other side: one only has to put oneself in the position of the advisors for the defendant companies, in the current prosecution, to see how the same technology might be of value to them. Whilst the companies will have voluntarily handed over huge volumes of data to the prosecuting authorities and will therefore have needed to understand that data for themselves, under the rules of criminal procedure in the UK, information unused by the prosecution which may afford a defence has to be disclosed by the prosecutor as well as the information the prosecutor is relying on. The defence will therefore need to understand not only its own corporate data but data obtained by the prosecutor from other sources at home and abroad and to understand that information in a context which may (or may not) afford some kind of defence to or mitigation of the charges.
One of the areas of investigation for both the prosecution and defence will be the communications between the company officers and the auditors and what exactly was misleading.
When the case is finally presented to a jury the vast volumes of data will have been skilfully narrowed to a minimum volume by the legal teams. We may learn exactly how that was done, but in all probability it will remain a secret. LM
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56 |
Page 57 |
Page 58 |
Page 59 |
Page 60 |
Page 61 |
Page 62 |
Page 63 |
Page 64 |
Page 65 |
Page 66 |
Page 67 |
Page 68 |
Page 69 |
Page 70 |
Page 71 |
Page 72 |
Page 73 |
Page 74 |
Page 75 |
Page 76 |
Page 77 |
Page 78 |
Page 79 |
Page 80 |
Page 81 |
Page 82 |
Page 83 |
Page 84 |
Page 85 |
Page 86 |
Page 87 |
Page 88 |
Page 89 |
Page 90 |
Page 91 |
Page 92 |
Page 93 |
Page 94 |
Page 95 |
Page 96 |
Page 97 |
Page 98 |
Page 99 |
Page 100 |
Page 101 |
Page 102 |
Page 103 |
Page 104 |
Page 105 |
Page 106 |
Page 107 |
Page 108 |
Page 109 |
Page 110 |
Page 111 |
Page 112 |
Page 113 |
Page 114 |
Page 115 |
Page 116 |
Page 117 |
Page 118 |
Page 119 |
Page 120 |
Page 121 |
Page 122 |
Page 123 |
Page 124 |
Page 125 |
Page 126 |
Page 127 |
Page 128 |
Page 129 |
Page 130 |
Page 131 |
Page 132 |
Page 133 |
Page 134 |
Page 135 |
Page 136 |
Page 137 |
Page 138 |
Page 139 |
Page 140