Question-Intimacy

Intro

question-intimacy is a package used to estimate the intimacy of questions. It is released with EMNLP 2020 paper Quantifying Intimacy in Language.

Install

Use pip

If pip is installed, question-intimacy could be installed directly from it:

pip3 install question-intimacy

Dependencies

python>=3.6.0
torch>=1.6.0
transformers >= 3.1.0
numpy
math
tqdm

Usage and Example

Notes: During your first usage, the package will download a model file automatically, which is about 500MB.

Construct the Predictor Object

>>> from question_intimacy.predict_intimacy import IntimacyEstimator
>>> inti = IntimacyEstimator()

Cuda is disabled by default, to allow GPU calculation, please use

>>> from question_intimacy.predict_intimacy import IntimacyEstimator
>>> inti = IntimacyEstimator(cuda=True)

predict

predict is the core method of this package, which takes a single text of a list of texts, and returns a list of raw values in [-1,1] (higher means more intimate, while lower means less).

# Predict intimacy for one question
>>> text = 'What is this movie ?''
>>> inti.predict(text,type='list')
-0.2737383

# Predict intimacy for a list of questions (less than a batch)
>>> text = ['What is this movie ?','Why do you hate me ?']
>>> inti.predict(text,type='list')
[-0.2737383, 0.3481976]

# Predict intimacy for a long list of questions
>>> text = [a long list of questions]
>>> inti.predict(text,type='long_list',tqdm=tqdm)
[-0.2737383, 0.3481976, ... ,-0.2737383, 0.3481976]

Contact

Jiaxin Pei (pedropei@umich.edu)