@@ -6,12 +6,12 @@ def dig_web(): twits = getTwits() photos = getFlickers() words = [] -# for twit in twits: -# twit.content_tags = extractKeyWords(twit) -# words.append(twit) - for photo in photos: - photo.content_tags = photo.tags - words.append(photo) + for twit in twits: + twit.content_tags = extractKeyWords(twit) + words.append(twit) + #for photo in photos: + # photo.content_tags = photo.tags + # words.append(photo) return words if __name__ == "__main__": analyser.py @@ -1,13 +1,23 @@ +CLUSTERS = {} + class Cluster: def __init__(self, tags=None): if tags is None: - tags = set() - self.tags = tags + tags = frozenset() + elif not isinstance(tags, frozenset): + tags = frozenset(tags) + self.tags = frozenset(tags) self.contextables = set() + CLUSTERS[self.tags] = self def add(self, *args): self.contextables.add(*args) + def populate(self, contextables): + for contextable in contextables: + if contextable.belongs_to(self): + self.add(contextable) + def __repr__(self): return "%s => %s" %(self.tags, self.contextables) @@ -24,26 +34,42 @@ class Cluster: return len(self.contextables) - def create_clusters(contextables, clusters, tags): - new_clusters = [ Cluster(cluster.tags.union(set([tag]))) for cluster in clusters - for tag in tags if tag not in cluster ] - - for contextable in contextables: - for cluster in new_clusters: - if contextable.belongs_to(cluster): - cluster.add(contextable) + print "Start creating new clusters" + new_clusters = [] + for cluster in clusters: + for tag in tags: + if tag not in cluster: + tag_cluster = CLUSTERS[frozenset([tag])] + new_cluster = Cluster(cluster.tags.union(tag_cluster.tags)) + new_cluster.populate(cluster.contextables.union(tag_cluster.contextables)) + new_clusters.append(new_cluster) + return list(sorted(new_clusters, key=lambda i: len(i))) +def merge_clusters(contextables, number=5): + print "Merging" + tags = order_tags(contextables)[:160] + new_clusters = [] + print "First clusters" + for tag in tags: + cluster = Cluster(frozenset([tag])) + cluster.populate(contextables) + new_clusters.append(cluster) + for i in range(number-1): + new_clusters = create_clusters(contextables, new_clusters, tags)[:160/(i+2)] return new_clusters -def merge_clusters(contextables, number=5): - tags = order_tags(contextables) - clusters = [ Cluster(set([tag])) for tag in tags ] - new_clusters = create_clusters(contextables, clusters, tags) +def order_clusters(clusters): + clusters = {} + for cluster in clusters: + for tag in contextable.content_tags: + tag = tag.strip() + if tags.has_key(tag): + tags[tag] += 1 + else: + tags[tag] = 1 + return sort_dict_by_value(tags) - for i in range(number-2): - new_clusters = create_clusters(contextables, new_clusters, tags) - return new_clusters def order_tags(contextables): tags = {} clusters.py @@ -1,4 +1,4 @@ from analyser import dig_web from clusters import merge_clusters -mc = merge_clusters(dig_web(), 2) +mc = merge_clusters(dig_web(), 5) main.py @@ -87,6 +87,17 @@ class TestCluster(unittest.TestCase): self.assertEquals(len(c), 27, c.contextables) + def test_php_clustering(self): + clusters = merge_clusters(ctxs, 1) + cluster = Cluster(set(["php"])) + for c in clusters: + if c == cluster: + break + else: + self.fail("Cluster not found!") + + self.assertEquals(len(c), 2, c.contextables) + def test_php_ruby_clustering(self): clusters = merge_clusters(ctxs, 2) cluster = Cluster(set(["php", "ruby"])) test_clusters.py 677b9068afae4ac29d6933c3c7e3015665c5c14b Bruno Gola bgola@pedal.(none) http://github.com/Fabs/eventomeeter/commit/2309475438f5f7c171a13402451abc3a39c08a5c 2309475438f5f7c171a13402451abc3a39c08a5c 2008-11-09T01:46:28-08:00 2008-11-09T01:46:28-08:00 estou com sono para colocar qualquer comentario que tenha algum sentido... ce07529e1329166b0b1ef6694d5aeca80c8520ed Bruno Gola bgola@pedal.(none)