The paper presents the results of a comparative analysis of two algorithms for cluster analysis "Lingo" and "STC". A corpus of documents obtained from the monitoring non-indexable sites specific thematic from "dark web". It is shown, that Lingo algorithm provided a better quality of clustering.