gensim's LSI model to calculate the similarity of documents

[python]  view plain copy

1. from gensim import corpora,models,similarities  
2. dictionary=corpora.Dictionary.load('/tmp/deerwester.dict')  
3. corpus=corpora.MmCorpus('/tmp/deerwester.mm')  
4. print (corpus)  
5.   
6. lsi=models.LsiModel(corpus,id2word=dictionary,num_topics=2)  
7. doc="human computer interaction"  
8. vec_bow=dictionary.doc2bow(doc.lower().split())  
9. vec_lsi = lsi [vec_bow]   #convert the query to LSI space  
10. print (vec_lsi)  
11.   
12. #transform corpus to space and index it  
13. index=similarities.MatrixSimilarity(lsi[corpus])  
14.   
15. index.save('/tmp/deerwester.index')  
16. sims = index [vec_lsi]  
17. sims=sorted(enumerate(sims),key=lambda item:-item[1])  
18.   
19. from  pprint  import  pprint  
20. pprint (sims)  

Using gensim's LSI model to calculate the similarity of documents