contexto_nlp: Automated Q&A for assessing lexicon acquisition
Use example:
In [2]: from contexto_nlp.nlp_core import profile_dicts
...:
...: up = profile_dicts(sources=["imdb", "twitter"], ctxt_user="user",
...: access_token="key", access_secret="pass", consumer_token="ckey",
...: consumer_secret="cpass")
In [3]: up.fit(["the lion king", "the matrix"])
Out[3]: <contexto_nlp.nlp_core.profile_dicts at 0x7f3cbc476358>
In [4]: plan = up.get_user_qa_plan() # See CSV files...
In [5]: plan
Out[5]:
{'the lion king': QA \
14 lion kingdom mini
2 lion kingdom mini
18 lion kingdom mini
...
answers length \
14 [(lion, 1.0), (kingdom, 1.69314718056), (mini,... 3
2 [(lion, 1.0), (kingdom, 1.69314718056), (mini,... 3
18 [(lion, 1.0), (kingdom, 1.69314718056), (mini,... 3
...
sentiment sim_wrt_topic \
14 {'neg': 0.0, 'neu': 1.0, 'pos': 0.0, 'compound... 0.992697
2 {'neg': 0.0, 'neu': 1.0, 'pos': 0.0, 'compound... 0.992741
18 {'neg': 0.0, 'neu': 1.0, 'pos': 0.0, 'compound... 0.903997
...
topic_rank
14 3
2 0
18 4
... ,
'the matrix': QA \
4 movieberto matrix
5 they spar again
1 neo laughs but is unsettled
...
answers length \
4 [(matrix, 1.0), (movieberto, 2.09861228867)] 2
5 [(spar, 2.09861228867)] 3
1 [(neo, 1.40546510811), (laughs, 2.09861228867)... 5
...
# Now you can get explicit questions and answers with difficulty 1 (0-3):
In [6]: up.pose_qa(save_posed=False, difficulty_sector=1, save_posed=True)
# See the corresponding qa_for_*.csv files