init globalproperties phrase extraction for each word (i=0 to n) for each word (j=i to n) check if phrase i to j exists if so get properties for each property, check if it exists if it exists, add 1 to the frequency else add to global properties Could use docker: dbpedia lookup http://lookup.dbpedia.org/api/search.asmx/KeywordSearch?QueryString=eggs dbpedia spotlight