When analysing text context, we can go beyond polarity (i.e. positive, negative and neutral as in sentiment analysis) and detect specific feelings and emotions (angry, happy, sad, concerned, excited, confused, etc.), this is called emotion analysis and is one of the important tasks under AI and Natural Language Processing (NLP).
Why emotion analysis
Emotion analysis is the process of identifying the underlying emotions expressed in textual data. This is beneficial to businesses to understand emotions expressed by their customers and staff and the reasons behind them.
There are various models and platforms providing emotion analysis to their customers. Most of these models have a fixed list of predefined emotions (ranging from 4 to 10).
The output might be one emotion or a list of emotions and their associated scores assigned by the model as shown in the image below.
We use neural network models for emotion analysis. These networks are trained on large datasets and vastly tested on in-domain and out-of-domain testsets.
We have provided you with a no-code platform in which you can apply state-of-the-art neural models to your data and benefit from the analysis.
Not only you can access overall emotion statistics in your dataset but also when combined with filtering you can analyse "emotion over time" and emotions in different categories (e.g. in a department, or a projects).
Below you can see two samples of emotion analysis on client feedback data on our dashboard. The first one illustrates the changes over time with three of the emotions (annoyance, pride, disappointment) highlighted.
The second one is the emotions within a cluster on reported "defects and issues" by clients.
Updated about 1 year ago