The document discusses sentiment analysis in social media. It makes three main points: 1. Human communication, both online and offline, is inherently subjective. Opinions and facts are often intertwined. 2. Facts and opinions expressed online have business value for consumers, marketers, competitors and managers. Sentiment analysis can turn attitudes expressed online into useful data. 3. Basic keyword searches and document-level analyses are not enough for decision support. Sentiment analysis requires techniques that can analyze features, opinion holders, semantics and propagate sentiment across multiple channels. It needs both machine and human capabilities.