GEDCOM search engine

The Idea

I imagine a simple site with a file upload form, which would allow anyone to register and upload any number of GEDCOM files. You wouldn't have to register to be able to upload, but to take advantage of the certain features of the site you'd need to do so. Registration would be free. The site could be supported by subscription fees for advanced features (but with still plenty of capability at a free level) or by targeted advertising via AdWords or the like.

Another part of the site would be a search form, that would allow me to specify basic genealogical criteria (name, surname, dates, places, etc) and search for individuals, marriages, places, or sources from the set of all GEDCOM files that have been uploaded. It would fun to explore some Ajaxy niceness. I can imagine a number of search form fields, which I could fill in press a button, and without a page refresh, narrow my search results accordingly.

After registering, the backend system, having processed my GEDCOM, would examine all the other GEDCOMS that had been submitted and provide a list of GEDCOM submitters I could contact that would be most likely to be involved in similar research.

Prior Art

This may already be implemented to some extent by One Great Family or Ancestry OneWorldTree, but I'd like to see something a bit more open. Both of the above feel too much like Walled Gardens. Rootsweb's WorldConnect gets us part of the way there, but this idea goes one step further, by making the service do much of the investigative work for me.

Key Benefits

• Being able to be put in contact with others doing similar research is the number one value I see a service like this providing.
• A secondary benefit is being able to find leads for pedigree data. Places like WorldConnect are often sparse on source data, but they do often provide a good startting point for further research.

Sample Search Result

Search results would be reminiscent of Google results, but with some genealogical twists:

John Smith
Born: 23 December, 1888 in provo, Utah, Utah, USA
Died: 1 January, 1943 in Provo, Utah, Utah, USA
Browse Pedigree - Download GEDCOM - Download GEDCOM of ancestors - Download GEDCOM of descendants - Etc...

Challenges

• Matching up related pedigrees could potentially be an O(n^2) algorithm. We'd have to find ways to make it smarter. A start might be to take the most common surname in my GEDCOM I've uploaded, and look for all the other GEDCOMS that have a high concentration of that same surname. I imagine the folks listed in the #Prior Art Section probably have algorithms they use to find and match similar lines.

I've decided to start working on this. It will take some time to implement, but the matching algorithms are not nearly as bad as O(n^2). I've had previous experience writing matching algorithms - they're not that difficult to write once you have training data, which the LDS Church makes available.--Dallan 19:42, 6 May 2006 (MDT)

• This type of system could be made obsolescent by some of the new centralized family tree databases in progess (such as the LDS Church's new Family Tree system).

I think a wiki approach to sharing pedigrees is a worthwhile enough idea that it deserves to be implemented.--Dallan 19:42, 6 May 2006 (MDT)