We study the problem of coding of unordered sets of words, appearing in natural language processing, retrieval, machine learning, computer vision, and other fields.  
We note that this problem is different from a problem of coding of a particular sequence of same words, and that considerable savings in rate may be achivable by allowing atrbitrary permutations of words in the decoded set.
We review several known results about this coding problems, and offer one possible design of codes that can be used for solving it. 
We show, that in memoryless model, and under certain conditions, our proposed code approaches the rate of
         $$ h T - \log m!  + O(m)$$
where $m$ is the number of words in the set, $T$ is the combined length of all words, and $h$ is the entropy of the source.