Rebecca Girigalla

Data mining and big data is what the digital industry runs on today. So far only structured data has been made use of for commercial activity. NARBS, short for narrative bits is a term coined by Prof Ananda Mitra, from Wake Forest University. He has analysed the use of posts and comments, also known as unstructured data, and successfully made prediction on peoples choices and behaviour. He discusses its uses and implications in this interview.

"This is a big data game" Courtesy: Wake Forest University

“This is a big data game”
Courtesy: Wake Forest University

“The main intention is to predict the behavior of individuals or groups based on thinking and understanding”, said Prof. Ananda Mitra on the mechanism of NARBS. NARBS, he said could supplement certain areas of research and has found applications in many industries.

Big data, NARBS and their uses today
On news agencies using NARBS in social networks he explained the importance of the presence a reporter. A single journalist can only gather so much a lot of people talking about an event will increase the volume of NARBS. “One has to then shift through NARBS to make sense of the information. It can become part of a main story – citizen reporters’ data then could be aggregated”, he explained.

“Through NARBS one can look at behavior and attitude. It is possible to parse information and gather details on events and opinions”, he said. This information comes from our personal posts on social networking website – and is sold to companies to analyse and be made use of. His take on the subject being the commodity is that it is the way the media economy is set up – to support news for the industry to exist.

Big data is now what advertising in the print industry used to represent – the mode is now different – instead of publishing, an advertisement that will be seen by millions and getting a response from hundreds, big data now allows institutions and corporations to analyze data and send out personalized messages to a selected audience with a higher chance of response?

“There is a bigger implication for example TOI selling ad space the only chance for an advertiser is to cast a large net and hope for a few responses. NARBS analysis allows us to throw a dart at a single individual”, said Prof. Mitra. The traditional advertising model, he felt, is going to disappear. One has only to look at the sidebar advertisements seen in platforms like Facebook which are displayed based on keywords from our activity on the website.

The shift from mass communication to personalised communication
We are used to advertising to a homogeneous audience. But this is no longer the case. Research shows that the audience is heterogeneous and fragmented”, elaborated Prof. Mitra. Institutions no longer think of a vast audience but of a single person. “That’s were data gets useful – it allows advertising to target a single person”, he added.

News, too is produced for a large audience but technology today allows for customization of news. This makes the job of a journalist difficult as they must address a smaller audience and hence must re-think the impact of their copy on the reader.

Owning our data
With big data, an aggregate of our personal information, being owned by  institutions, Prof. Mitra explains that there are two organizations that own data – the government  and the corporations.
If the government owns data, then questions such as our freedom being at at stake come into play. “Being law abiding doesn’t stop one from thinking critically. Before you choose, they would know what you would choose through big data and NARBS.  And if they don’t link the choice, they would have the power to stop an individual from his choice by either persuading or oppressing”, said Prof. Mitra.

The second kind of institution that owns personal data, the corporations, compete for attention vis a vis ones consumption of products. Prof. Mitra’s point of view was that if a corporation has personal information – being a few markets – they could provide one with targeted information so that we would buy – not oppressive but manipulative.

Should big data allowed?
From a moral point of view, Prof. Mitra debated, on whether the use of big data used. From a citizens perspective, he felt, it did not seem positive in a democratic system. But it is allowed legally. A conniving democratic government could ensure that they would not be voted out. It is here that the ethical issue gets complex and interesting.  Competing parties might block each other out as the targeted audience would be bombarded with messages from both sides and based on those would make their choices. This is increasingly relevant with the kind of research we are talking about.

Scale of operations
“If we are dealing with 100 people, and through predictive we find 10 people  who might need what we advertise – should half of those people respond, it would mean a 5% sale rate”, explained Prof. Mitra. If scaled to a million – it would mean 5000 prospective people to target without advertising to a million through a newspaper or other traditional media through a system that is far less expensive.

While the accuracy is low, the volume is high and would not work unless the audience is high. “This is a big data game”, he added, “ It is basically finding the diamond in the rough. The chance is now higher”.

Based on voices of individuals as the data, the accuracy cannot go very high. Other tools with higher accuracy such as a questionnaire with specific questions would achieve higher accuracy.

People not needing a product – we can go on the presumption that certain people need something and with big data our past purchases are already known. Knowing more of your increases the level of accuracy and a fairly accurate prediction could be made.

On errors made in prediction
Predictive policing, like we have seen in the movie – Minority Report, could be really dangerous. Because mistakes could be made and people would be harmed. “It is worrisome”, he felt.  Countries that are paranoid could go to great lengths and worry about the use of information for crime prevention and profiling.

Every government does profiling. Profiling will appear to be more detailed with NARBS. But as of today we are not fully aware if it would be accurate or not. One has to have a responsible law enforcement that would recognize the accuracy level.

Big data making us less human?
Being human has been called in to question . Here one has automatically given weight to the old human. “Perhaps being human is in both our analog and our digital presence”, mused Prof. Mitra. We meet people physically then see them on Facebook and then feel that we know the person. He explained that it is in that combination lies our identity.