Predata relies on web metadata - the data about the data - to capture overall browsing habits rather than individual expressions. We’ve found that patterns of behavior replicate more reliably in the metadata than in the data itself. For this reason, the Predata platform can generate predictive signals that are impossible to reveal from sentiment analysis, natural language processing, or other forms of content analysis. Predata doesn’t try to analyze everything. The data is hand-curated based on our models of human behavior, which means the system only uses data relevant to the questions you want answered.