You can start with the following points to understand, in building out a similar site.

Collection of web logs

Using data mining technology, the site user access to the web server in the process of log data generated by the analysis of processing, so as to find out the web user's access patterns and interests and hobbies.

Integration of data collection, cleaning, storage, dimensional analysis, data visualization of this series of functions of data storage and analysis platform.

Generally speaking, the massive data storage in order to facilitate the future use of big data data processing will use hbase to store.

Collected data storage can first use relational databases such as mysql to do, and then the processed data will be pressed into the hbase.

As for tagging and foreground visual query it is better to use nosql database for distributed processing, such as mongodb.