Several industry commentators are predicting that 2015 will see a rise in investment and advancements made in the area of investigative analytics. Understanding the key challenges facing any organisation when monitoring, deterring, and investigating fraud and crime will prove vital in the investigative analytics investment paying off.
The influence of Big Data across the organisation is seeing an increased awareness of the main pain points – too much data for some systems to handle, messy data, the need for network and visual analysis, and efficient case management.
As crime and fraud gets more sophisticated, there is more pressure on investigative analytics solutions to keep up and deliver value. With the end goal in mind, it is important to effectively manage the input data that will affect the quality of the final result.
Effective processes are needed to address shortfalls in data like misspellings, incompleteness or even data in the wrong place. Data enrichment will add to usefulness and relationships to other data, thus increasing its value. Then, accurate, automated and intelligent linking/matching is necessary across all data sets to enable regular and visual analytics supporting investigations, policy and/or prosecution.
White collar, corporate crimes are influenced by a range of factors and understanding them will aid investigative analytics efforts. They include:
Sophisticated technology – fraud and crime are carried out via various mediums for different purposes. From money laundering using virtual currencies like Bitcoin to misusing CCTV footage and compromising data and organisational security, advances in technology can be countered by forming an effective case through the underlying data. Facilitating collection of disparate information, extraction of entities, augmenting it with relevant meta-data -boosting data sets, establishing and inferring links and uncovering informal networks through categorisation of data are essential in managing such schemes and practices.
Diversity - both geographic and linguistic diversity such as physically different offices, telecommuting from multiple locations and the influence of local language add to the complexity challenging investigations. First line of defence technology like access control and detection now needs to be substantially boosted when preventing fraud or in preparing a prosecution case. Techniques like network and content analytics, both facilitated by effective data standardisation and entity matching solutions will prove to be highly beneficial. Investigations can be assisted by appending geographic and geo-demographic data to help identify locations of organisational assets, the proximity to the employee or customer and exploring other relationships between them. Similarly, the influence of a multinational work force often requires the need to perform advanced lexical analysis – such as by using a library of name and keyword aliases in uncovering incriminating evidence in transactions and communications.
The 3Vs of Big Data – the Velocity, Volume and Variability associated with company information makes it decidedly harder to use for investigative purposes. The frequency and amount of information shared, especially in unstructured formats increases the challenges presented when trying to mine data sets for implicating information. High performance linking and analytical engines are essential in establishing entities and associations between them, as are the use of several techniques like natural language processing and text analytics. Parsing unstructured text frequently involves deciphering colloquial language, trends like tagging comments (not just within social media posts) with hashtags (#), emoticons and various other forms of media. Rule based categorisation of related information will be an essential pre-requisite for visualisation and case management.
Group psychology – corporate crimes often involve a group of people working together to fraudulently conduct business. Cartels may operate subtly or there may be open threats by groups to conduct criminal activities. These types of operations are increasingly coming into organisations’ radars.
Uncovering plans, identifying patterns, tracing activities they engaged in and establishing timelines in which they were conducted are once again challenged by the problems posed by Big Data. Semantic analysis is an important technique that can be used to understand concepts and the links in the information exchange by these groups. Sentiment analysis can then be used to establish a positive or negative connotation in conversations.
Most fraudsters operate with the intention of never being caught so their efforts in covering up their tracks are also increasing in complexity. From deliberately planting erroneous ‘evidence’ to creating false leads and dummy information with the aim of leading investigators down the wrong path, organisations are preparing themselves by creating effective, reliable and comprehensive investigations systems.
As the industry matures and more organisations adopt an ‘intelligence repository’, the case for data quality management by feeding and maintaining clean data will only grow.
Article written by Darren Wu.
Darren is a sales manager at Intech Solutions with over 15 years working with data, data systems and solving business issues. For a comprehensive insight into the role data quality plays in investigative analytics, call the Intech Solutions team today on 61 2 8305 2100 or email sales and let us guide you through the foundations for setting up an effective intelligence repository.