FACT360: eDiscovery in Action
)
This Article originally appears on FACT360's website, read it here.
Professor Mark Bishop,
FACT360, Chief Scientific Adviser
mark.bishop@fact360.co
FACT360’s Chief Scientific Adviser, Professor Mark Bishop brings together the theory discussed in the ‘The Science of AI’ series (read the first post here) explaining how it can be applied to organisations today and why what people ‘do’ can be more insightful than what they ‘say’…
Imagine this…events unfold, and you are dropped into the opening of a long and complex Fraud and Corporate Threat investigation with 600,000 emails and documents to sift through, and you’re not even sure who you are looking for, what you are looking for or when any incidents of interest may have occurred. Well, perhaps surprisingly, the underlying problem isn’t new and historically has engaged some of the world’s brightest minds – Plato’s for one. In 380BC Plato framed the following Socratic dialogue, called ‘The Meno’:
“How do you enquire Socrates into that which you know not?”
“What will you put forth as the subject of enquiry?”
“And if you find what you want, how will you ever know that this is what you did not know?”
‘Meno’s Paradox’, as it became known, can be reframed as follows:
If you know what you’re looking for, inquiry is unnecessary. If you don’t know what you’re looking for, inquiry is impossible. Therefore, inquiry is either unnecessary or impossible.
Meno’s Paradox, Plato 380BCBecause corporate investigators typically begin with ‘that which we know not’ (i.e., they don’t know how a threat developed, was realised, and evolved), the Meno directly relates to the problem of investigating novel Fraud and Corporate Threat today.
Plato’s response to the paradox is found in his “Theory of Recollection” wherein, the process of knowledge exploration and discovery is simply cached out as the ‘recollection of timeless forms, from a period long before our immortal souls were imprisoned inside our physical bodies’. Slightly more prosaically, at FACT360 we believe the answer to Plato’s paradox can be revealed through a combination of Transactional Analytics and Natural Language Processing.
At Bletchley Park during the Second World War, Alan Turing – BBC’s “Icon of the century” – was primarily concerned with decrypting communications; decoding message semantics (e.g., `Commander A requests permission from Commander B to move troops from Bayeux to Caen’), whereas his colleague Gordon Welchman was primarily concerned with communication interactions; decoding message transactions (e.g., `Commander A in Bayeux, communicated with Commander B in Caen, at 9:17am on the morning of June 6th, 1944’). It is interesting to note that, unlike Turing’s work on breaking the Enigma codes, much of Welchman’s research still falls under the Official Secrets Act.
In this context, at a meeting of the European Council on Foreign Relations in London on the 19th April, 2018, Claudia Aradau – Professor of International Politics in the Department of War Studies, King’s College London – specifically highlighted how social transactions – what people actually do – often reveals more useful intelligence than what people say. The metaphor Claudia used to highlight this contrasted the ‘little lies we might tell on a Facebook or dating site profile, with the transactional fact that Person A communicated to Person B, at a certain time and place.
Inspired by Gordon Welchman’s Traffic Analysis of Axis signals, ‘Transactional Analytics’ leverages human social interactions across communication networks, to detect subtle changes in behaviour characteristic of potential covert activity. Fundamentally misconstruing what Transactional Analytics offers, the CEO of a major UK data security company once revealed to me that he would “never communicate anything `too sensitive’ in corporate communications”. But, because `Transactional Analytics’ reveals `changes in communication transactions’, and not message semantics, the particular topics being discussed in communications don’t impact the intelligence it reveals.
So, how does FACT360 leverage Welchman’s insight? Using emails (and, as available, other linguistic records: phone transcripts, chatroom data, minutes of meetings etc), FACT360 can graph the communication flows within an organization as an interconnected network of nodes; wherein each node represents an employee (identified by their email addresses, telephone numbers etc), with directed arcs linking to communicating nodes and contacts.
The underlying processing architecture of FACT360 processes communications in the temporal domain; converting discrete events (i.e., sending and receiving groups of discrete symbols; email messages; documents etc) into streams of continuous multivariate data. Subsequently, by automatically highlighting anomalies in these signals, FACT360 reveals ‘unknown unknowns’ from data, i.e. the things you didn’t even know that you didn’t know, foregrounding potentially critical behavioural change. Thus, by combining ‘transactional analysis’ with state-of-art `semantically tuned’ NLP algorithms (e.g., identifying topics and concepts of interest within communications), FACT360 can dynamically flag:
- Changes in employee `prestige’ (e.g., a lower-ranking employee now communicating with a ‘Member of the Board’), indicative of promotion or, vice versa, an employee dropping ‘out of the loop’ in corporate strategies.
- Communications on ‘concepts of interest’, correlated with known time-line events (potentially linking to known covert activity).
- The ‘sentiment’ of communication; the use of sexist or bullying language etc. (indicative of workplace harassment).
- Team members withdrawing (signalling lack of engagement).
- Company-wide team dissipation (suggestive of disintegrating corporate structure).
At FACT360, we believe intentions are reflected in behaviour, and behaviour is reflected in use of words. Thus, because words represent ‘topics of discussion’, points in time where such concepts appear (or disappear) can indicate a change-point in our behaviour, because people’s motivations change. However, as outlined in previous blogs, programming machines to genuinely understand the meaning of texts (Natural Language Understanding) is difficult, if not impossible.
So, at FACT360, in revealing critical anomalies and documents of interest to investigators, we predicate what people `do’ – via communication transactions – not just what people `say’. Or to put it another, it is eDiscovery in action.
Read Part 8 here – “The Mechanics of Computation: Automata”
Professor Mark Bishop is FACT360’s Chief Scientific Adviser and to see how these leading-edge scientific techniques can be applied in your organisation download our White Paper “The Science of FACT360”or get in touch info@fact360.co.
This Article originally appears on FACT360's website, and you can read it here.