Numeric sentiment prediction

[MaxGit095]

I am currently performing a sentiment analysis using PyABSA, using the following code:

''' 
You can check this library: https://github.com/yangheng95/PyABSA 
I am using the latest version 2.0
'''
from pyabsa import ABSAInstruction, available_checkpoints

myreviews = ['I really enjoyed the place, e.g. the service and food were great','Nice and clean place, would go again.', 'I love the chef, he makes the best pasta']
    
checkpoint_map = available_checkpoints()
quadruple_extractor = ABSAInstruction.ABSAGenerator("multilingual")
quadruple_result = quadruple_extractor.predict(review, max_length=128)

However, my result does currently look like this:
{'text': 'I really enjoyed the place, e.g. the service and food were great', 'Quadruples': [{'aspect': 'service', 'polarity': 'positive', 'opinion': 'enjoyed', 'category': 'SERVICE#GENERAL'}, {'aspect': 'food', 'polarity': 'positive', 'opinion': 'great', 'category': 'FOOD#QUALITY'}]}

My problem is that I need the sentiment (polarity) as a figure, but I can't find any description in the documentation.

[fawern]

In PyABSA, sentiment polarities are typically represented as categorical labels such as 'positive', 'negative', or 'neutral'. The framework does not natively assign numerical scores to these polarities. However, you can map these categorical labels to numerical values based on your specific requirements. For instance, you might assign 'positive' to 1, 'neutral' to 0, and 'negative' to -1.

[fawern]

from pyabsa import ABSAInstruction, available_checkpoints

myreviews = [
    'I really enjoyed the place, e.g. the service and food were great',
    'Nice and clean place, would go again.',
    'I love the chef, he makes the best pasta'
]

quadruple_extractor = ABSAInstruction.ABSAGenerator("multilingual")

polarity_mapping = {'positive': 1, 'neutral': 0, 'negative': -1}

for review in myreviews:
    quadruple_result = quadruple_extractor.predict(review, max_length=128)
    for quadruple in quadruple_result['Quadruples']:
        aspect = quadruple['aspect']
        polarity = quadruple['polarity']
        opinion = quadruple['opinion']
        category = quadruple['category']
        # Map the polarity to a numerical value
        numerical_polarity = polarity_mapping.get(polarity, None)
        print(f"Aspect: {aspect}, Polarity: {polarity} ({numerical_polarity}), Opinion: {opinion}, Category: {category}")