Besides sentiment polarity that indicates positive, negative, and neutral types of attitudes to the described topic, Content Analysis API can discern emotional state. A more advanced approach gives the opportunity to detect the following range of emotional reactions:
- anger
- happiness
- love
- sadness
- share (desire to share content)
- fun
Content Analysis API relies on machine learning (ML) algorithms that programmatically determine the emotional tone of a certain text. To give such results they’re trained on vast volumes of carefully selected data. Our model is implemented on the Hybrid method. The algorithm processes certain pre-labeled datasets and learns from them until it can independently identify the sentiments of previously unknown pieces of content. At the same time, we can always manually modify the sentiment detection rules which ensures the highest possible accuracy of our product.