The easiest way to emulate is this process is through a numerical scale. Take into account social indicators such as personality type (A,B), age, affluence, beliefs, values, perspectives, and most importantly what you are looking for in the other person. These scores are then "coded" by attributing a number to each answer which is then added or subtracted depending upon response. These values ultimately create a number and that number is "matched" with similar numbers (matrix matching) and you can layer this process to be more precise.
In terms of your project: trying to pair up entrepreneurial minded people. I would start with a specific geographic location, Say... the Bay area. Right there the parameters are defined to specific coordinates (numeric). I would rate everyone on a proximity rating with each other first (which i am assuming that you are trying to do with the 1 city limitation) people that are with in 10 mile of each other receive a rating of (10), 15 miles (9)... 50+ miles (1) ect. The logic that drives this the proximity principle: objects that are close together will be seen as belonging together, so people that live in close proximity to each other will perceive the other to belong. Next I would create an in-depth survey to try to generate as much relevant information such as skill sets, interests, beliefs, values, age, what they are looking for in a "match" and also they type of company that they are trying to build. If it is extremely technical then there should be 2 very technical co-founders, if less technical then I would say 1 technical & 1 semi-technical co-founders. Depending upon what needs the project has. I would then match people who have "complimentary" scores. I would then generate scores using my coded survey and match the people together.
In terms of your project, I think it would be a good idea to go through different online survey tests (match, zoosk, eharmony) to see how the questions are set up and then coded.
(Here is a link to good coding practices. I think it will be able to help you further understand how you generate a numerical score off a survey) http://books.google.com/books?id=dB0fw5lf0GQC&pg=PA345&#...