|
| Predicting social media sentiment with Setu |
| K Rajani Kanth / Chennai/ Hyderabad Sep 23, 2011, 00:33 IST |
|
Enterprises typically want to stay tuned into buzz around their brands, products or services, especially when it concerns social media chatter. They also like to engage themselves with customers, understand their perceptions, develop insights and use the feedback positively.
Setu Software Systems, an information technology start-up incubated at the International Institute of Information Technology-Hyderabad, claims its social media analytics platform provides enterprises a solution to perform these tasks.
In an already-cluttered industry, there are some very popular sentiment analysis tools. For instance, WiseWindow's sentiment analysis tools provide an index of social media sites, blogs, and message boards, and feed predictive business intelligence models, while SAS, a business intelligence and analytic services provider, offers a new social media monitoring tool. Wildfire, a US-based start-up focused on creating interactive promotions on social sites, recently launched a social media monitor—a tool that tracks and compares how brands perform on Facebook and Twitter. Other popular tools include Radian, Techrigy and Alterian.
So, what makes Setu's social media analytics platform unique? “Typically, extracting user insight stops at the product level. Ploughing the feedback from social media at the specific attribute level, the time dimension—how the product or service changes on the first or the third day, the second week or the third month (of the purchase or implementation)—and spatial aspects like what people in various countries think about the product is very difficult. That's when companies would get the right pulse about their technologies, products and services,” says K Vasudeva Varma, co-founder and chief scientist, Setu.
“The strength of our social media analytics platform is capturing the attribute level of such opinion and feedback. It is equally good at understanding multiple languages, over 70 world languages, which enables it know what the Dutch are talking about the product in Dutch, and what the Germans are talking about a service in German,” he says.
Setu has built its own proprietary Setooz crawlers to scoop millions of posts from at least 30 very popular social media networks, blogs, news articles, reviews, discussion forums, image and video-sharing sites. The technology has the capability to parse a wide variety of documents and process large datasets, using its indexing and searching mechanisms.
Varma says the Setu Sentiment Analyser is the platform's core engine, built on the cloud middleware, which measures the buzz of the customer to predict the sentiment. It is an agglomeration of both statistical and rule-based models, and predicts sentiment with high-accuracy. “Once the information is gathered and analysed, the platform presents it through a pie chart. The insights are reported visually in the form of charts and graphs to the end user. The platform has the support to have e-mail alerts for generated reports, export them to different formats and view them, based on attributes such as region and gender. Right now, we are deploying the platform within enterprises by plugging it into their servers. In the future, we are looking at offering it to consumers, and this should happen in a year,” he says.
Yet another innovation from Setu's stable is an automatic content creation technology that generates structured, unambiguous information, ready for machine processing, from unstructured, ambiguous text written in human languages. “For instance, if there are 10,000 cellphone models available in the market, human experts write reviews comparing a few popular phones, like LG versus Samsung, and so on. They can write a maximum of 50 to 100 articles. However, the auto content creation tool on our Compare.setooz.com can write 10,000 x 10,000 articles automatically, and in a fraction of a second. All these articles would be very detailed,” Varma adds.
Currently, the platform generates articles only in English. However, it can also generate articles in 70 world languages. “The platform generates articles on the fly when you key in words and click the search button. Primarily, we take the authoritative information (from manufacturers and social media) and then rely on the redundancy that is available on the web, which means it is important. Right now, only cellphone information is available on the platform and we are looking at adding all the consumer electronics information on it,” Varma says, adding Setu is looking at filing for 10 patents for these technologies.
|
|
|
|
|
|
|
|
|
|
 Read Business news in |  |
|
|
|
|
|
|
Advertisements |
|
|
|
|
|
|
|
|
|
|
Content Guide
