def topics_from_to(start, end): years = range(start, end+1) topicFiles = fileSys.traverseTopicDirecotry(model_path, 1, years) # fill the path where distances are stored clf_topic = path.join(root_path, 'dtm/distance') distanceFiles = fileSys.traverseDistanceDirectory(clf_topic, years) topic_graph = graph.createGraph(topicFiles, distanceFiles, 0) return topic_graph
def topics_for_class(class_mode, class_name, start, end): if class_mode == 'acm-class': clf_topic = ioFile.load_object(path.join(root_path, 'class_topic/class_topic_acm-class.pkl')) elif class_mode == 'arxiv-category': clf_topic = ioFile.load_object(path.join(root_path, 'class_topic/class_topic_arxiv-category.pkl')) clf_topic_stat = [] topic_num = [] years = set(range(start, end+1)) for year in range(start, end+1): try: clf_topic_stat.append(Counter(clf_topic[str(year)][class_name])) topic_num.append(len(set(clf_topic[str(year)][class_name]))) except KeyError: years.remove(year) topicFiles = fileSys.traverseTopicDirecotry(model_path, 1, years) clf_topic = path.join(root_path, 'class_topic/distance') distanceFiles = fileSys.traverseDistanceDirectory(clf_topic, list(years)) topic_graph = graph.createGraph(topicFiles, distanceFiles, 2, clf_topic_stat, topic_num) return topic_graph, years