home.markdown

@@ -11,7 +11,9 @@ Horst is an acronym for 'holistic recommender and storytelling technology' and i
 **Key features:**
 
 * In-memory graph representation with small memory footprint. We e.g. host the whole music area in a ~1GB RAM graph at [Horst](http://horst.dfki.de). This graph comprises currently(year 2020) ~21 million nodes with ~67 million connecting relationships.
-* Fast multi steiner tree implementation, offering nearly-cheapest (cheapest if you calculate to the end), connecting subgraphs between a set of graph nodes. In the case you give just two query nodes, you will receive the connecting paths. If all relations and nodes have the same weight, this implements a shortest path algorithm - but even faster than other implementations we tried (maybe you had more luck).
+* Fast multi steiner tree implementation, offering nearly-cheapest (cheapest if you calculate to the end), connecting subgraphs between a set of graph nodes. In the case you give just two start nodes, you will receive the connecting paths. In the simple case where all relations and nodes have the same weight, this implements a shortest path algorithm - but even faster than other implementations we tried (maybe you had more luck).
+* Node recommendations by specifying start nodes and result node types.
+* RDF module to load arbitrary rdf graphs and offer the Horst functionality on top of it.
 
 **Try it with the whole music domain at [Horst](http://horst.dfki.de)!**