This is one of the more interesting talks this year: Georg Heeg got involved in a project to find the Bach House in Koethen. Apparently, the commonly identified house isn't the one Bach actually used when he was in that city.

Not the kind of project you would normally associate with a software company, but that's the kind of thing Georg gets himself involved in. They didn't use the standard historical approach that has been used - they figured that's been tried and failed. So... they used an exclusion algorithm to rule places out, and then focus on what was left. That involved a lot of data collection

First, they had to collect all the extant data, and get it into softcopy (OCR, hand entry, etc, etc). Then they created a semantic networ with searching functionality. There are products out there - k-Infinity, atlas-ti (too many manual steps). So - they created a system from scratch to handle the problem.

The project was built in VW with Seaside, and used Gemstone for the backing store. They borrowed the Tries search code from the Shootout winner code. To do this, they extended the meta model in VW (interestingly, not possible in Gemstone - so they had to do transforms in and out).

The basic problem was this: addresses on notes to and from Bach contained names, titles, and professions, but not street addresses - seems that the people delivering the mail knew the locations by that information. To make things more complex, Bach did not own a house in the city, or rent from the local Prince - instead, he rented from someone else.

So what was known? He had a big family, and many students living with him (i.e, a big home). That allowed some exclusion right off. In going through the possibilities, size and records of when the homes were built allowed more exclusions. One of the homes they excluded is the one that historians have identified as the probable home. That left three homes, all owned at the time by the same person.

One was a shop back then, so it was eliminated. One more piece of information - Bach apparently complained of noise from a nearby water mill. So it seemed that they had found the place that Back lived in briefly - That left one set of houses just outside the old city walls, and it is known that wealthy citizens lived in that area. The houses there had been excluded based on the notion that they were built too late (1729) - but they were actually built starting in 1719. Also, the homes were tax exempt until all lots were filled in - and it's known that taxation started in 1730. By piecing this together with other stuff I'm not fully transcribing, they were able to narrow it down to one house, which was built in 1719.

Great talk - I think this one will come across much better when I release the video.

Technorati Tags: smalltalk, smalltalk solutions