Old Voter List — Search India's 2002 Electoral Rolls for SIR

Search across scripts

Type in English. Find a name printed in Bengali.

A state's roll is printed in that state's own script. You don't have to be able to type it. Every name is indexed twice — as printed, and romanized — so a Latin-keyboard query reaches a Bengali or Devanagari row and still scores it honestly.

Search results for a Latin-script query, showing matched names printed in Bengali script with match scores beside them. — Rampurhat (AC291), West Bengal — names blurred here, not on the site.

Where the rows come from

A 2002 page, scanned. Read line by line.

Much of what a CEO office publishes is a photograph of paper — no text layer, nothing to select or copy, decades of shelf on it. Those pages go through OCR tuned per state, then a parser that knows what a roll row looks like. Every result on the search slide was lifted from a page like this one.

A scanned page of the 2002 West Bengal electoral roll, typewritten in Bengali, with numbered elector rows. — The source page for part 75 — scanned, not typed.

What we could not read

When we can't read something, we say so.

“The roll did not record an age for you” and “we could not read the age the roll printed” are different answers to the question you came here with, and only one of them is ours to assert. So a field we failed to read says exactly that, on the row, beside the name.

we could not read an age here
sex not read
name not read: row begins at the relation word

The same rule runs the other way: a search filter that cannot match anything tells you why instead of returning a bare zero.

Coverage

22 of India's 36 states and union territories. The rolls are public in every one of them.

Indexed so far: Arunachal Pradesh, Assam, Chandigarh, Chhattisgarh, Delhi, Goa, Haryana, Himachal Pradesh, Karnataka, Lakshadweep, Madhya Pradesh, Meghalaya, Mizoram, Nagaland, Puducherry, Punjab, Rajasthan, Sikkim, Telangana, Tripura, Uttarakhand, West Bengal — 2018 constituencies.

Every state and union territory published a roll for this era, and all of it is public data. What a state needs in order to appear here is a connector — the code that turns its particular file format, font or scan quality into rows this index can search. That is essentially the whole gap between a state being covered and not.

These figures are read from the running index on each page load rather than typed into the page, so they can be quoted as they stand — the per-constituency breakdown is here →

How this gets built

Built by AvaDhuri, with the encouragement and support of officials at the Greater Bengaluru Authority.

AvaDhuri builds tools for working with data in ways that can be checked rather than taken on trust; this one applies that instinct to a real, public dataset. It exists because of generous guidance from officials at the Greater Bengaluru Authority, and continues to be developed with their support. It remains an independent project — not affiliated with the Election Commission or any state CEO office.

The most useful thing anyone can contribute is a state. The pipeline is open source — open a pull request with a connector for a state we don't cover and we'll review it, build the index, and put it live. Running the servers costs money too, and a contribution helps, but it is the second-most useful thing.