I'm working on a project similar to Klout which provides detail about how you influence other people and who influenced you. We'll be fetching data from few social networking sites (i.e linked in, facebook, twitter etc) to analyze how users interacts with one another. For that we need to parse the data and store it in db and have to analyze it so that strength of relation of two user can be decided. We'll be accessing data offline as well to provide them with accurate results.

If we consider facebook activities, we need to have access to Facebook users' news feed, wall data which includes likes,comments,shares etc. To decide how one user influence other, we'll store all the data and analyze it.

I need suggestions on what steps need to be taken for great performance. We'll be using ASP.Net(C#) Web forms, SQL Server, jQuery. Main concern is parsing of data, it's storage and retrieval with least overhead. For that I've summarized few points as below :

• Should we switch over to document-oriented database, like MongoDB or RavenDB for the whole app or part of it even though none of team member have experience with them?
• Should we use SQL Server Analysis service?
• Is there any other library than Json.NET for parsing data?
• Is it advisable to use any C# library over FQL + GET Request ?

I've tried to provide as much info as possible. Please share your views for the same.