Why is it better to examine emotions in a text instead of sentiment?

When sentiment analysis began to be used for marketing activities in the early 2000s, it opened up vast opportunities for the marketing and advertising industries. Understanding consumer sentiment allowed for better validation of actions and more precise targeting of target groups. Today, 20 years later, new doors are opening—rather than just measuring sentiment, we can now analyze emotions, and this offers even more benefits. What are they?

Differences Between Sentiment and Emotions: Full List of Emotions

First, let’s define sentiment and emotions—the most important differences arise at this basic level because sentiment and emotions are two distinct phenomena.

• • Sentiment is the state in which a person experiencing emotions can name what they are feeling, connect that experience with thoughts about it, and then make a conscious decision regarding the source of the stimulus. Sentiment, therefore, is the sum of physiological reactions (from the body) and cognitive processes triggered by the experience of emotions. Since it is a conscious state, it can be sustained by the person experiencing it. In general, sentiment is a broad mental attitude towards a particular experience.

• Emotions, on the other hand, are the physiological and, consequently, psychological brain responses to an external stimulus or an experience related to such a stimulus. The combination of the experience and the body’s reaction (hormonal activity) causes a short-lived and unconscious state, leading to specific actions such as fight, flight, freeze, awe, disgust, etc. What’s more—each emotion is tied to a different type of reaction, resulting in different behaviors.

As you can see, without experiencing emotions, you cannot enter a state of sentiment. Emotions operate at the lowest, behavioral level, while sentiment is their reflection, consideration, and evaluation by the person experiencing them. Emotions cause spontaneous reactions, while sentiment leads to conscious and controlled actions. The most significant difference: there are only eight basic emotions (according to Plutchik’s theory: joy, sadness, anger, fear, disgust, surprise, anticipation, and trust), which, when combined, form secondary, higher emotions that lead to the creation of feelings. In the case of sentiment, we are dealing only with positive and negative sentiment, sometimes also neutral.

Exploring Emotions: Media Monitoring vs. Sentitool – A Tool Comparison

Just as the phenomena of sentiment and emotions differ, so do the tools used to analyze them. These tools leverage advanced technologies such as neurolinguistic programming, machine learning, and other AI algorithms. Currently, in the Polish market for sentiment analysis in online content, several reputable companies exist, but in the field of emotion analysis, only one stands out as a true pioneer—Sentimenti. Below is a comparison of the general capabilities of sentiment and emotion analysis tools.

Conclusion: The common (even within the industry) usage of the terms sentiment analysis and emotion analysis is incorrect. Sentiment is a much narrower concept, indicating in practice only the tone of a statement and possibly the mood of the author. Emotion analysis, on the other hand, describes the level of individual emotions (providing a percentage result for eight components, as well as the type of sentiment and the level of emotional arousal); with this data, consumer reactions and behaviors can be predicted with high accuracy.

As you can see, this data is more detailed; its analysis is more challenging, but it is also more accurate and comprehensive. Sentimenti’s algorithm includes as many as 30,000 words and phrases collected from a group of 22,000 people. The algorithm itself was developed in collaboration with the Wrocław University of Science and Technology and the Brain Imaging Lab of the Polish Academy of Sciences.

How Do the Results of Both Algorithms Look in Practice?

Time for some practice! Take a look at the following analysis of a real review:

I will never buy anything from smelly V(…) again. I recently decided I didn’t want to buy tragic clothes from chain stores, so I bought from there, and the supposedly perfect condition item, without signs of wear, had a stain that was visible. The woman claims that when she sent it, the stain wasn’t there.

This review is authentic and taken directly from a portal. What can you read from it when measuring sentiment versus emotion?

Results: Anger – 70%, Fear – 45%, Anticipation – 24%, Surprise – 57%, Trust – 14%, Sadness – 60%, Disgust – 62%, Joy – 10%, Positive sentiment – 10%, Negative sentiment – 70%, Emotional arousal – 72%.

The comment comes from a disappointed customer who, instead of making a purchase from one of the popular boutiques, opted for what she considered higher-quality used clothes. The purchase turned out to be unsuccessful, and the item was stained despite being described as in good condition. The buyer is clearly upset (emotional arousal at 72%), angry (a high 70%), feeling disgust (62%) and sadness (60%)—as she expected a good purchase. Notice that the woman also exhibits a high level of surprise (57%) and fear (45%)—when you add the disgust result, you can understand that this is her physical reaction to the experience of being deceived.

Now look at the same review from a sentiment analysis perspective: you receive two results—negative (70%) and positive (10%). The negative sentiment is evident even without automatic analysis, given phrases like smelly, tragic, stain, woman. The phrases supposedly perfect condition and without signs of wear build the low positive sentiment score.

The key information, however, is hidden—the phrase smelly refers directly to the shopping platform, tragic to the quality of chain store clothes, and woman to the seller. The overall tone of the review is set by the epithets unrelated directly to the purchased item, yet the sentiment analysis result remains negative.

Conclusion: With percentage results for basic emotions and knowledge of the typical physiological reactions they correspond to, you can estimate consumer behavior in this situation. Anger is associated with an attack response (hence the comment on the portal), disgust and fear with flight, and sadness with freezing. The emotional consumer will likely switch to another platform or opt for in-person shopping. She will probably not use this platform again.

Emotion Analysis is Effective Beyond Content Marketing

You’ve just seen how sentiment and emotion analysis results are interpreted in a specific example, and what kind of insights you can gain from them. As you’ve likely noticed, emotion analysis is a far more comprehensive, complete solution tailored to customer needs. But are emotion analysis algorithms limited to marketing, PR, or customer service?

Definitely not. With the development of machine learning technology and the implementation of increasingly sophisticated AI algorithms, the possibilities of emotion analysis extend into other industries. Today, emotions can be analyzed, for example, to forecast stock market prices or investment opportunities in the cryptocurrency market.

Future Prospects for Emotion Analysis Tools

If the above information hasn’t convinced you of the superiority of emotion analysis over sentiment analysis, look at the development potential of the former. Artificial intelligence is already highly advanced. Today, AI is being implemented not only to study but also to create emotionally engaging content aimed at achieving set goals.

These goals include increasing conversion rates from marketing activities, acquiring more effective leads, or providing customer service tailored to consumer needs (including potential crisis situations and ways to avoid them). The latter goal significantly improves user experience, increasing customer loyalty to the brand, which, in turn, strengthens the brand’s position in the market. That’s the potential of this technology—sentiment analysis can’t do that.