A news story in Hindu of 06 Jan 2015 states “on an experiment, about 58,000 probable duplicate entries in Chennai were spotted. Subsequently, the conventional system of field verification was carried out. Eventually, 15,736 entries were found to be duplicate. “That’s how we removed them from the rolls,” explains Ajay Yadav, Joint CEO (Information Technology)”
A CEO-TN press release states, “2,87,598 names were deleted due to shifting, death of electors and duplicate entries. Also, on experimental basis a photographic deduplication
software was tried for Chennai Corporation. It resulted in detection of 15736 probable duplicates.”
Header line in Hindu states “Novel exercise to spot duplicate voters.” and the byline “Authorities embark on complex initiative.”
De-duplication is an activity mandated by ECI before publishing every new version of electoral rolls. It is neither novel nor complex. ECI conducted a workshop of IT officials from CEO’s offices of all states on 24th and 25th May, 2013, at Nainital. They were kind to invite me. In the conference, CEO, Manipur, talked about how she identified duplicate voters by comparing photographs of voters using features of Picasa from Google.
I have been reporting suspected duplicate entries in the rolls of various states based on voter record text comparison. Inspired by the talk by CEO, Manipur, I added photo comparison feature to my software. ECI helped me by sending a small sample database dump of Chattisgarh to prove the feature. In Jul to Sep 2013 I demonstrated the software to ECI technical team. The software was installed at ECI and proven. Subsequently, I guess that the software was not used.
From 04 Jul 2014 to 26 Aug 2014 officers of CEO-TN (including the Joint CEO) and I exchanged 16 emails. I got a small sample dump of their voter database to run my application to detect duplicates by comparing voter record text, followed by photo comparison. I shared the application with them and helped them run it on their computers. Then there was silence from their end.
Whereas CEO-KA has been responding to my observations stating that, “We know the errors in our ERMS and know how to correct them” and do nothing for years, I am glad that CEO-TN has acted to correct the errors in Chennai rolls. New version of the rolls are published on 05 Jan 2015, which I shall analyse to verify the quality.
I also hope that other CEOs take lessons from CEO-TN though they have mostly ignored my input for years.