A method of automatic weighting in retrieval where an algebraic model is used for term frequency and distribution, creating representative vectors in multiple dimensional space; when compared to the vectors of an incoming query, the relevance of results may be predicted