[D] How to deal with paragraphs in opinion mining (Sentiment Analysis)?

My goal is to find comments about stocks and create a model to classify the comment as good/bad/neutral (or maybe a 5 point scale so I can have good, really good, etc.) so that I can give an opinion for the stock. My issue is that I have some comments with multiple sentences. In some cases, all sentences are about one stock and in other cases, the sentences are about different stocks.

A simple solution to this could occur when …

analysis datascience deal mining opinion sentiment analysis