lovelyai999 commited on
Commit
8c3133c
1 Parent(s): 5396799

Update myEmbedding.py

Browse files
Files changed (1) hide show
  1. myEmbedding.py +0 -61
myEmbedding.py CHANGED
@@ -574,64 +574,3 @@ EmbeddingConcept =lambda x :ConceptNetwork.embeds(x).tolist()[0]
574
 
575
 
576
 
577
- def updateWiki():
578
- import random
579
- from fileDict3 import FileDict
580
- wiki=FileDict(Gbase+"wikipedia.sql3")
581
- wikiConcept=FileDict(Gbase +"wikipediaConceptA.sql3")
582
- random.seed(time.time())
583
-
584
- n=800*100*100
585
- startIndex=len(list(wikiConcept.keys()))
586
-
587
- #f1=open("testBuildTree.txt" , "w" ,encoding="utf-8")
588
- #f2=open("testConceptTreeSentence.txt" , "w" ,encoding="utf-8")
589
- m=0
590
- for x ,(k,v) in enumerate(wiki.items()):
591
- if x <startIndex:continue
592
- #k,v=next(items)
593
- #if random.randint(1,3000)!=10 :continue
594
- #m+=1
595
- #if m>1000 :break
596
- if x%5==0 and IN_COLAB:google.colab.output.clear()
597
- its=[]
598
- text=k +"\n"+v
599
- t0=time.time ()
600
- print(Text.Cyan(k))
601
- try:
602
- conceptTree=ConceptNetwork.generate_concept_tree(text,split_sentences=True,top_n=0)
603
-
604
- s2=format_concept_tree(tree=conceptTree,max_depth=16)
605
- print (s2[:2048])
606
- wikiConcept[k]=s2
607
- #json.dumps (conceptTree)
608
- except:
609
- print (v)
610
- import traceback
611
- traceback.print_exc()
612
- t=time.time ()-t0
613
- print(Text.Cyan(x) ,Text.LightYellow(t))
614
- if x%100000==0 and x>startIndex+90000:
615
- wikiConcept._commit()
616
- wikiConcept.close()
617
- drive.flush_and_unmount()
618
- drive.mount('/gdrive',force_remount=True)
619
- return 1
620
- return 0
621
-
622
-
623
- #f1.write(k+":\n"+s1+"\n")
624
- #f2.write(k+":\n"+s2+"\n")
625
- #f1.close()
626
- #f2.close()
627
-
628
-
629
-
630
-
631
- if __name__ == '__main__' :
632
- while True :
633
- if not updateWiki():break
634
-
635
-
636
-
637
-
 
574
 
575
 
576