You need to expantiate a bit on what you are trying to do. Co-occurrence of two words might be used as an indication of similarity; LSA is based on this logic. I'm not sure of what co-occurrence of documents and words mean?
Co-occurrence of words with documents can also be used as a measure for semantic similarity. This was done, for example, in Explicit Semantic Analysis (ESA), http://www.cs.technion.ac.il/~gabr/resources/code/esa/esa.html