Training data for a recommender system of a location based social network

I'm currently developing a location based social network in Ruby on Rails. I also want to include a recommendation system. For testing the algorithms of this recommendations I need some real, anonymous training data. I've found the data from the Netflix Prize, but they are only including .

I'm searching for data that includes

• users
• friendships
• locations or venues
• check-ins (like in foursquare)

Does anybody know a good source for such data? Or a 开发者_开发问答proven algorithm for generating this data? Or any other idea?

---

Search for random graph generation algorithms (more prciese, "social graph generation") to simulate social graph. Try retrieving the some test geolocation data by Google maps API or similar services. Unfortunately, I don't know what is "check-ins (like in foursquare)".

Also see Free Social Graph Data

---

I've finally solved it by using the gowalla API. Here you get a lot of information about users, without asking the users to permit the access to their data (kinda strange, but it works).

check it out: http://gowalla.com/api/explorer#/users/sco