This is the term for decompiling a huge database in order to discover the relevance of documents by comparing them to other documents. This is done through stemming, local and global weighting and normalisation.