The first approach is based on the standard information
retrieval heuristics such as topic frequency and inverse document
frequency weighting and stopword removal. The downside
of this approach is that it does not directly take into
the account the user behavior.