Sounds like just the sort of task to throw some machine learning at, call it BiasBot and train it on all ideological angles - no matter whether you happen to agree or disagree, the thing needs to know about all angles - and let is loose on whatever source you want to check for bias.