Apr 25, 2015

Making sense of a million responses

Share on
  • Published in:

A Robert Jerard Ravi, advisor (quality of services) at the Telecom Regulatory Authority of India (Trai), must be worried. When his official email id, advqos@trai.gov.in, was mentioned on the consultation paper on the regulatory framework for over-the-top services for receiving public comments, little did he anticipate the flood of mails he would get through the next month. On Thursday, savetheinternet.in, the platform collating the responses of netizens on the issue, claimed public responses on the matter had crossed the one-million mark.

Ravi isn't alone. Earlier this month, the Ministry of Corporate Affairs sought three months from the Bombay High Court to pass a final order on the merger of Financial Technologies with National Spot Exchange Ltd, as it had received about 19,000 representations from Financial Technologies shareholders.

About a year ago, the same ministry had to deal with about 60,000 responses on the new companies Act.

Other regulators such as the Securities and Exchange Board of India, the Insurance Regulatory and Development Authority of India regularly seek public views on their proposals. In several cases, it is mandated that such proposals are put before the public.

As more and more respondents take to emails and dedicated public response platforms such as savetheinternet.in and change.org, processing such a large number of responses is fast becoming a concern, especially as stakeholders insist no response can be taken lightly. To make things more difficult, Trai's consultation paper, which received a record 49,462 hits on the website, ran into 120 pages and sought public views on 20 questions.

"I'm not sure how Trai will deal with this but we hope each of the million citizens who mailed Trai is heard and his/her view is taken on board and respected," savetheinternet.in volunteer Nikhil Pahwa told Business Standard.

"We hope this sets a precedent in terms of public participation in regulatory and government consultations in India, and that future consultations are presented to the public in an accessible and bipartisan manner, with tools such as savetheinternet.in to assist in participation."

Preethi Herman, country lead and campaigns director, Change.org India, says, "Now, it is easier for citizens to make their voices heard than at any point in history. This has huge positive implications for the Indian democracy. It's time institutions took online democratic engagement more seriously. In the coming months, we will be rolling out even better tools to bridge the gap between decision makers and citizens through our site – we're seeking to be the number one platform for people to engage with the institutions that govern their lives on a huge range of issues."

Last year, the corporate affairs ministry had commissioned a platform to receive responses on the hundreds of sections and sub-sections of the Companies Act. The platform, built by Corporate Professionals, allowed section-wise responses; it classified responses under different heads such as drafting errors and conceptual issues. Further, separate log-in ids were provided for different sections of stakeholders. This organised collection of responses helped the ministry compile the data in a matter of two-three days, says Pavan Kumar Vijay, managing director, Corporate Professionals.

While establishing a dedicated platform to receive comments with stock responses can be a solution, big data experts suggest using natural language processing (NLP) tools and opinion mining techniques.

Recently, some big data experts at LIRNEasia, a Colombo-based information and communication technology think-tank, brainstormed on the issue. A paper posted by Sriganesh Lokanathan, team leader (big data research, LIRNEasia, discusses possible solutions, including NLP tools such as word clouds and semantic analysis. "Individuals or groups responding to questions might have specific interests or motives in taking part in the debate. They might use words/phrases relevant to such interests or motivations consistently across questions, though not necessarily frequently enough to be noticed in the word clouds for each question. A word cloud that visualises the aggregate of all responses to all questions for which words/phrases have been excluded might serve to highlight underlying patterns of interest of the respondents," the paper said.

Simple text matching and probability measures based on semantic similarity could be used to identify responses identical or derived from a common source, it added.

Another useful indicator could be the quality of the language used in the response. The quality of writing could be used as a filtering mechanism to identify responses that might be worth greater scrutiny for navigating a complex subject such as net neutrality.

However, Lokanathan says while these techniques could simplify the work to an extent, "human intuition and intervention are needed for effective interpretation".

Request a Call