The way we write says a lot about our intentions. “Sentiment analysis” is predicated on the idea that writing – whether in an email, text or memo to a colleague, for example – can be generally defined in one of three ways: positive, negative or neutral. It uses algorithms that can scan large amounts of text, looking for keywords and measuring tone much more quickly, and efficiently, than could a person.
The applications for this kind of technology are nearly endless. Companies can scan social media messages to learn what about their business or products is trending, whether it be in a negative or positive way. Marketers can measure keywords to discover buying tendencies and analyze their audience demographic.
Sentiment Analysis for Internal Use
Employers can (and have) used it to monitor employees, including one case (cited by an article in Harvard Business Review) in which “semantic analysis identified a small team of salespeople in the middle of negotiating their defection to a competitor. The sentiment analytics software had identified both atypical frequency and vocabulary between the sales people and – more provocatively – radically different exchanges between the sales people and key accounts.”
That last example should spark a light bulb in every fraud investigator’s head. When analyzing employees’ communication, there are keywords and language that can tip companies off that their employers are engaging in fraud. It can be an invaluable tool that raises red flags, hopefully before a fraud scheme is so far developed that it is costly and crippling to the company.
History of Sentiment Analysis
The concept isn’t all that new: The New York Times reported in 2009 on sentiment analysis and its potential. In “Mining the Web for Feelings, Not Facts,” Alex Wright explained:
An emerging field known as sentiment analysis is taking shape around one of the computer world’s unexplored frontiers: translating the vagaries of human emotion into hard data.
This is more than just an interesting programming exercise. For many businesses, online opinion has turned into a kind of virtual currency that can make or break a product in the marketplace.
Yet many companies struggle to make sense of the caterwaul of complaints and compliments that now swirl around their products online. As sentiment analysis tools begin to take shape, they could not only help businesses improve their bottom lines, but also eventually transform the experience of searching for information online.
Even back then, as the NYT reported, an industry was emerging and several companies were already offering subscription services to provide sentiment analysis – though it was geared chiefly toward companies measuring sentiment among buyers, not their own employees.
Test Case: Enron
However, researchers have already been able to test sentiment analysis within the fraud realm in a backward-looking way. Using Enron emails (that were made public during the scandal), sentiment analysis revealed a series of red flags throughout the lifetime of the fraud: “the red marks flag emails where the sender's sentiment suddenly turned sharply negative (and would therefore be a good place to start looking for evidence).” Read more and see the graph.
Right now, sentiment analysis still seems to be used only on a limited basis. However, we may one day see a time when it is a standard internal control, and a part of most fraud prevention strategies. Sentiment analysis won’t prevent every fraud, nor help close every case. But it can provide a good starting point to knowing that something is amiss.