This study addresses the challenges of extractive summarization in opinionated online forum threads, highlighting that traditional text summarization methods may need adjustments due to the unique nature of user-generated content. The research involves creating an annotated corpus to explore summarization techniques that consider relevance, text quality, and subjectivity. Key features identified for effective summarization include sentence length and the number of sentiment words, which were found to be highly correlated with good summary sentences through unpaired Student's t-test. The study proposes modifications to a standard Integer Linear Programming-based summarization framework to incorporate these features.