Yes, this is available as standard in Insights, available straight after your survey has closed. The analysis provides you both themes and sentiment. You can also easily – from within the Application – download the text data for review elsewhere, for example in a spreadsheet.
How accurate is Qlearsite’s sentiment and theme analysis on open text comments?
Very accurate, our models are trained to find the most appropriate themes in a collection of sentences using the words, sentence structure and many other characteristics. Our models go beyond simple text classification, so called ‘bag of words’ models, often used for word frequencies and sentiment analysis.
Unfortunately, no NLP model based on machine learning is ever 100% accurate as they are based on probabilities. Should the tool not be able to recognise the sentiment of a comment than it will fall back on connecting specific words to themes. Over the longer term, we look to balance the model’s aggressiveness in finding themes, versus the level of accuracy we demand when it thinks it has found the most suitable theme. Too aggressive and the accuracy suffers, too accurate and the model’s usefulness suffers. Finding this balance improves with time, more data and training from us.
We’ve spent considerable time over the last few years, training the models on a wide variety of texts. But when the model is faced with a context or phrase it has not seen before, there will always be a chance that it gets the theme wrong. The chance of this happening continues to diminish over time with more and more data.