Thank you for all pablo. I'm waiting for the end of the downloading, and
reading about your suggestions. 

At the end, the database will be for a private usage (academic research), so
the number of users is limited to volunteers and colleagues collaborating in
the project.

Our aim is to take a portion of the past, and coming from a point zero follow
the story and see how qualitative data (from interviews, blogs, rss, etc.) and
quantitative data (from stats and attributes collected) can explain the
performance life cycle of the different entities studied (players, teams,
nations...).

We know where how to find this portion of past and how to collect (thanks to
you) the data in scattered html pages generated from a database from a server.
If i understand well, there, serious people will go to sql data management
system. So the last problem, in the collection of data is to do the
translation from my gross data to my net data in sql format that i'll can
exploit with the ibm academic software. Maybe, it will appear stupid but what
i have to do for that? I am open to read everything around the subject.