(self,pnode,anomaly_cutoff)
| 415 | # cnt += 1 |
| 416 | return g |
| 417 | def caculate_anomaly_score(self,pnode,anomaly_cutoff): |
| 418 | need_to_caculate = defaultdict(list) |
| 419 | # nodes = self.G.nodes() |
| 420 | # have better way to find neighbor? |
| 421 | undirected_G = self.G.to_undirected(as_view=True) |
| 422 | print(len(pnode)) |
| 423 | for node in pnode: |
| 424 | neibor = list(undirected_G[node]) |
| 425 | # print(node) |
| 426 | if len(neibor) == 0: |
| 427 | continue |
| 428 | # tmp_dict = set() |
| 429 | for nei in neibor: |
| 430 | |
| 431 | if self.GetNodeName(nei) is not None and self.GetNodeName(nei) != 'unknown' and self.GetNodeType(nei) != APTLOG_NODE_TYPE.PROCESS: |
| 432 | need_to_caculate[node].append((nei,self.GetNodeName(nei))) |
| 433 | # tmp_dict.add(self.GetNodeAttr(nei)) |
| 434 | # except Exception as e: |
| 435 | # print(nei,self.GetNodeAttr(nei)) |
| 436 | print('need_to_caculate',len(need_to_caculate)) |
| 437 | |
| 438 | if len(need_to_caculate) == 0: |
| 439 | return -1 |
| 440 | node_feature = defaultdict(list) |
| 441 | |
| 442 | for node in need_to_caculate.keys(): |
| 443 | flag = False |
| 444 | cmdline = self.GetNodeName(node) |
| 445 | split_path = sanitize_string(cmdline) |
| 446 | if len(split_path) == 0: |
| 447 | continue |
| 448 | new_name = '/'.join(split_path) |
| 449 | self.nodes[node]['newname'] = new_name |
| 450 | tmp = [] |
| 451 | for l,i in enumerate(split_path): |
| 452 | tmp += [self.c2v.wv[i]] |
| 453 | r = np.mean(tmp,axis=0) * self.mean_tfidf |
| 454 | |
| 455 | node_feature[node] += [r.tolist()] |
| 456 | tmp_dict = set() |
| 457 | process_objects = set(need_to_caculate[node]) |
| 458 | for id,object_name in process_objects: |
| 459 | split_path = sanitize_string(object_name) |
| 460 | if len(split_path) == 0: |
| 461 | continue |
| 462 | flag = True |
| 463 | new_name = '/'.join(split_path) |
| 464 | self.nodes[id]['newname'] = new_name |
| 465 | |
| 466 | tmp = [] |
| 467 | for l,i in enumerate(split_path): |
| 468 | tmp += [self.w2v.wv[i]] |
| 469 | r = np.mean(tmp,axis=0) |
| 470 | try: |
| 471 | t = self.tfidf[new_name.lower()] |
| 472 | except: |
| 473 | t = self.mean_tfidf |
| 474 | r = r * t |
no test coverage detected