I have tweets obtained based on matches (football) before the match begins. I have tweets which specify a team will win 3-1 and so on which are easily analyzed using regular expressions. I am facing difficulties when it comes to tweets where two team names are specified with win / loss comparison such as:
Manchester United will win the match tonight against Spurs. (or) Spurs will find it difficult to break down United defence.
I have taken several such tweets and replaced the team names as TEAMONE and TEAMTWO and placed them into two files (teamone winning tweets and team two winning tweets). Using SVM, the model couldn't clearly classify tweets between the two classes (files mentioned above). Is it due to class imbalance? Should I stick to a rule based classification in this case or would any different method work? I'm a newbie in ML and any suggestions would be appreciated.