Thursday, March 11, 2010

The Algorithm

In 2004, I developed a simple normalization algorithm that made it very easy for people to search multilingual databases, especially those with Devanagari text. It was called SouIndics (after Soundex) and was a major reason why a search engine built by a wonderful team worked so well.

Today I received a mail asking for details of that algorithm. So I went searching. I was surprised to find it was still being used in 2007 and maybe even later! Even more surprising was the fact that there were companies using it!

One company is a direct descendant of the organization that built the original search engine, and they seem to have moved on rightly to other things.

Another (BinyasIT, iSearch, Google cache) however touts it as:

Exclusive SouIndics Phonetic matching: similar sounding word indigenous technology.

I don't know what's exclusive about a piece of GPL-licensed code, but clearly marketing has to show it's doing some work. I am, however, incredibly grateful that they kept the original name. As recent events in the movie industry demonstrate, not many follow the principle of credit where credit is due. Of course, I can be extremely cynical and say that SouIndics is a brand name, and people better say things like they "support SouIndics" if they want to be taken seriously in the multilingual search space, but that would be wishful thinking.

2 comments:

Alff said...

I still remember this as the project where you worked with a professor from IIT and had to do something with searching the electoral list. Too bad I am not great with details.

ksp said...

Yep, those were fun times. I doubt this algorithm would seen so much exposure if it weren't for him (incidentally, he was from VJTI, though his alma mater was IIT) and his team.

I'm surprised you remembered this!