| from transformers import pipeline | |
| classifier = pipeline("text-classification", | |
| model='bhadresh-savani/distilbert-base-uncased-emotion', | |
| return_all_scores=True) | |
| def get_emotion(text='No text yet'): | |
| prediction = classifier(text)[0] | |
| result = sorted(prediction, key=lambda x: x['score'])[::-1] | |
| return result | |
| sentiment_map = {'anger': 'neg', 'sadness': 'neg', 'fear': 'neg', | |
| 'joy': 'pos', 'love': 'pos', 'surprise': 'pos'} | |
| good_arcs = ['neg - pos', 'pos - neg'] | |
| great_arcs = ['pos - neg - pos', 'neg - pos - neg'] | |
| def get_sentiment_arc_evaluation(emotions): | |
| sentiment_arc = [] | |
| for emo in emotions: | |
| sentiment = sentiment_map[emo] | |
| if sentiment_arc and sentiment_arc[-1] == sentiment: | |
| continue | |
| sentiment_arc.append(sentiment) | |
| sentiment_arc_str = ' - '.join(sentiment_arc) | |
| if sentiment_arc_str in great_arcs: | |
| return 'What a great plot! Excellent! 😍' | |
| elif sentiment_arc_str in good_arcs: | |
| return 'Story plot seems nice! But you can do better. 😉' | |
| elif len(sentiment_arc) < 2: | |
| return "No judgment, but... The plot might be too simple! 🤓" | |
| else: | |
| return "The plot seems complicated. 🤔 But maybe I am just too stupid to understand!" | |