Using Latent Dirichlet Allocation with Gensim

Question

I am working on a project and I would like to use Latent Dirichlet Allocation in order to extract topics from a large amount of articles.

My code is this:

import gensim
import csv
import json
import glob
from gensim import corpora, models
from nltk.corpus import stopwords
from nltk.tokenize import RegexpTokenizer
from time import gmtime, strftime

tokenizer = RegexpTokenizer(r'\w+')
cachedStopWords = set(stopwords.words("english"))
body = []
processed = []

with open('/…/file.json') as j:
    data = json.load(j)

for i in range(0,len(data)):
    body.append(data[i]['text'].lower())

for entry in body:
    row = tokenizer.tokenize(entry)
    processed.append([word for word in row if word not in cachedStopWords])

dictionary = corpora.Dictionary(processed)
corpus = [dictionary.doc2bow(text) for text in processed]
lda = gensim.models.ldamodel.LdaModel(corpus, id2word=dictionary, num_topics=50, update_every=1, passes=1)
topics = lda.show_topics(num_topics=50, num_words=8)

other_doc = "After being jailed for life in 1964, Nelson Mandela became a worldwide symbol of resistance to apartheid. But his opposition to racism began many years before."
print lda[other_doc]

Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-  packages/gensim/models/ldamodel.py", line 714, in __getitem__
gamma, _ = self.inference([bow])
File "/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site
packages/gensim/models/ldamodel.py", line 361, in inference ids = [id for id, _ in doc]
ValueError: need more than 1 value to unpack

I also tried to use LdaMulticore in 3 different ways :

lda = gensim.models.LdaMulticore(corpus, id2word=dictionary, num_topics=100, workers=3)
lda = gensim.models.ldamodel.LdaMulticore(corpus, id2word=dictionary, num_topics=100, workers=3)
lda = models.LdaMulticore(corpus, id2word=dictionary, num_topics=100, workers=3)

And every time I got this error :

Traceback (most recent call last):
File "<stdin>", line 1, in <module>
AttributeError: 'module' object has no attribute ‘LdaMulticore'

Any ideas?

Thank you in advance.

Answer 1

You have to convert back to the phase space.

http://radimrehurek.com/gensim/tut3.html#similarity-interface

vec_bow = dictionary.doc2bow(other_doc.lower().split())
vec_lsi = lda[vec_bow] # convert the query to LSI space

Answer 2

I realize this is old, but I just had this same problem. You are probably pointing to an older version of Gensim. You have to make sure you're using version >= 0.10.2.

Update with "easy_install -U gensim" and then make sure your IDE is seeing the updated library.