PS: There is no summarized version, so sorry …! and map = frequency map not the game map…
Assumption: This topic is not currently involved in the research. I tried looking up for it in various forums and google searches but couldn’t find anything similar.
Problem I faced: This is with the reference to recent TI8-Dota2 casting. So in the casting there were four parallel channels going on so at a time one can focus on only one match. They tried manually to address the interesting parts of one stream into other but it was all messed up; changing suddenly from one interesting fight in a match to other.
Solution Proposal: Just merely by analyzing the frequency of the live comments that are being posted we can determine the intensity of the fight going on, but then the question arises how can we decide the threshold of the limiting frequency; simple train it. There had been thousands of game recordings online and as well as their summarized version (basically important or fucked up part of matches), we can look at those parts (summarized) and map them to original video and then create a whole new map of the frequency signal of the comments and can differentiate entire match’s frequency map. Further refining can be done based on sentiment analysis of the comments.
User Interface: Displaying multiple video frames at the point of time when such phenomenon occurs. The time limit for such phenomenon can also be found out by the map itself
Bi-products:
-
The trained map can be used as a extension for the streamers to summarize their match recordings, they do it manually currently.
-
I always wanted to create an Alexa skill that can me tell me that the match is going interesting (while I am doing some other stuff) so one may not miss it, but ended up getting fucked up by the APIs and resources limitation.
Problem in the proposal:
-
Different games and streams will have different map so deciding which streams to merge in the frames can be really complex.
-
To find out the minimum time for which the phenomenon(comment frequency change) should happen is a really complex task in itself.
-
I forgot this one but will list soon.
Stakeholders: Everyone watching the such streams, streamers, you developers and me 
Why: My professor asked me to write a paper/doc regarding a software related problem that you face/faced and how will you encounter it, so if the idea seems good to you guys I am more than happy; however I need to markout the software quality part of it but will see.