Saturday, May 16, 2009

If I'm perfectly honest, this exercise has been more painful than anticipated.

We started out by planning to publish a "skinny" version of every record in our CORE (COrroborated, Remediated, Enriched) Avox database. Then we realized that a good chunk of these records had been verified over a year ago for clients wanting just a one time clean-up (some folks never learn...) so those records had a reasonable chance of being out of date. As it takes a good 10 minutes per record to recheck, trolling through hundreds of thousands of stale records would take too long and cost too much for this initiative. Given that we are making this content publicly available so everyone can look at it, we thought it would be risky to publish data we knew had a high probably of being out of date.

Then in the pared down database, we found a large number of records that were not unique "legal" entities. That is to say, they were branches, departments or funds which had been included for specific Avox clients at their request. So while these are valid records from their perspective, we've excluded them from wiki-data for now with a view to starting with pure legal entities.

Then we released the database to a small trial audience and, as expected, we received helpful feedback on some quality issues which we have been in the process of addressing. This is exactly what we had hoped for as every new set of eyes that looks at the content brings with them new perspective and knowledge. The community helps improve its own asset. We also received very helpful feedback from users including some of our own staff about the layout of the search facility, the results and the issue filing process.

Perhaps most interesting and encouraging was the fact that one of our competitors provided constructive feedback. Could it be that we may be able to work with eachother to jointly improve our mutual clients' data? Let's see.

So now we feel the database, although slimmed down quite a lot, is ready for broad public viewing and consumption. There will be out of date information. There may be some duplicate values. And yes, I expect there will be some outright errors. It's all expected. All we ask is that you tell us about any problems you find. You just need to click on the "Correction" button next to the record in question.

I know it's a bit of a pain however please do share your feedback/thoughts with respect to wiki-data on this blog. You can just email me if you prefer. We also have a LinkedIn group (Avox business entity discussion forum) set up.