This answer is necessarily vague because
- my understanding of the article is a bit limited due to the wonky translation (Google Translate is pretty good but not perfect)
- this kind of government agency does not readily make detailed information available (for obvious reasons)
Mass surveillance of phone networks is easy. The government uses legislation, bribery, thuggery, … to get their surveillance equipment or software inserted into the network of the relevant phone network operators.
This easily yields both call content (what was said) and call metadata (who called whom, where the two parties were at the time of the call, when the call was made, …).
Mass surveillance of internet-based communication is harder. The government uses (as above) … inserted into the network of the relevant ISP.
This easily yields some communication metadata (what IP address communicated with what IP address, when, how big, how long) and yields some additional metadata where the metadata is unencrypted and yields some additional metadata through traffic analysis and yields some communication content where the content is unencrypted.
Well, that should start the discussion off anyway …