How Does NewsBrain Predict Interest?
When NewsBrain fetches an article, it predicts your level of interest in it. This consists of two major steps. First, it scans the article for relevant topics. Then it makes its prediction based on topic relevancy, and interest levels in the relevant topics.
The article is scanned, looking for topics that are in the Interest Profile. When it finds one, it calculates how relevant that topic is to the article. This involves how often the topic appears, where in the article it appears, and so on. It does this for each topic it finds in the article.
To make its interest level prediction, it looks at each relevant topic, the level of relevancy, and your interest in that topic. These are combined to make the final prediction.
When you look at the article, if you make a correction to this prediction, your correction propagates back through the Interest Profile, adjusting the relevant topics. This is how NewsBrain learns.