def directed_graph(self) : if not hasattr(self, "_directed_graph") : print "getting directed graph ..." graph = defaultdict(_dd_int) # Zhu: in my VM, build speed is about 1.4w entity / s ac = AcoraBuilder(*self.database.entities).build() # match consumes no time, compared to build for text, attrib in self.database : entities = zip(*longest_match(ac.finditer(text)))[0] for entity in set(entities) : if entity == attrib["title"] : continue graph[attrib["title"]][entity] += 1 delattr(self, "database") self._directed_graph = graph return self._directed_graph
