In addition to jakebeal's answer, let me address the "How to propose that change?" question.
Surely the requirement to have work indexed at Scimago and Scopus was instantiated to have some "measurable" guideline -- the initiators of the project probably wanted to avoid any discussions on what papers should count and what shouldn't later. While this is understandable, it has the problems of (1) giving people an incentive to game the system, and (2) keep people away from supporting new publication venues that are not yet on a list.
So all you can do is to have them make the list of acceptable publication venues as inclusive as reasonably possible - and you will need to follow their "visibility in terms of research" line of thought to have strong arguments. Here is one idea for doing so, although it means a lot of work.
Check how many venues that are not already in Scimago and Scopus have been used to present research that was mentioned in connection with the Turing medal. Surely, if venues that had such an impact are missing, your request for making the list more inclusive would be reasonable.
There is a CORE conference ranking - Quite a lot of effort has been put into assembling it and it is or was used somehow for research funding allocation in Australia (someone else may know more about that). While such rankings are not unproblematic (and this one is based on the Australian perspective), publications in conferences ranked "A" in this list should surely count. If you find some conferences on the list that are ranked A and are not indexed in Scopus or Scimago but are indexed on DBLP, this makes a reasonably strong case for the inclusion of DBLP.
You may also need to provide evidence that conference papers are so important for computer science. I think that publications on this matter have been mentioned here at academia.stackexchange, so you should also be able to find them.
On a related note, I'm not sure if all IEEE Xplore publications are indexed by DBLP. There used to be quite a significant lag between IEEE Xplore publication and addition to DBLP (>6 months), which is very late if the data is to be used for scholarship allocation. Also, IEEE Xplore used to contain proceedings of some "spamferences", but they may have kicked them out in the meantime.