Introduction

Aspect-based emotion analysis is a tool for finding the emotions in a text. It investigates the content, and checks what this text of information is about. It can be about products, brands, or other subjects; then, the analysis moves onto finding the prevalent emotions.

The most important ability of it is categorizing this emotional data, highlighting which information occurs more than once or in high intensity, thus understanding what needs improvement. Moreover, it allows us to examine the data from various angles, revealing trends such as the most favored aspects or areas of concern.

At the end of this analysis, information on how people feel about things like "Company," "Product Quality," or "Social Media” are gathered. This is an impressive way to understand people's real emotions and afterwards their reactions.

ABSA vs ABEA

Aspect-Based Sentiment Analysis (ABSA) has long been a cornerstone in understanding consumer sentiment, categorizing opinions as positive, neutral, or negative. While effective, ABSA has its limitations, particularly in its somewhat simplistic emotional categorization. This is where Aspect-Based Emotion Analysis (ABEA) comes into play, offering a more nuanced approach to data interpretation.

Like ABSA, ABEA starts by dissecting the text to identify various elements and aspects. However, it goes a step further by categorizing these aspects into specific emotional states (Alqaryouti et al., 2019). This process, known as emotion detection, extends beyond text to include audio and video content, providing a more comprehensive understanding of human emotions (Varni et al., 2020, cited in Cortal et al., 2022).

Our decision to build our product around ABEA stems from its ability to offer a more detailed emotional landscape. The tool sifts through text and categorizes it into predefined emotional states, allowing for a deeper, more nuanced understanding of consumer sentiment.

But this isn't just a data exercise. It's a targeted approach to align textual sentiments with specific entities, such as a brand or product features. By doing so, we can pinpoint exactly what delights or disappoints our customers, offering actionable insights for improvement. Some current uses for ABSA, and possible uses for ABEA are listed below:

Uses for ABSA and ABEA

1. Monitoring customer satisfaction levels with a specific product or service.

2. Monitor public opinion of companies (yours and your competitors).

3. Understanding how employees feel about their job and company. (Chiusano, 2022)

4. Track brand and product images.

5. Evaluate consumer experiences.

6. Perform market research. (De et al., 2022)

Once we've organized all this text, it opens up lots of possibilities. You can pick a keyword, an emotion, or both, and see what people are saying. This helps you find the most common opinions and feelings. With the classification and categorization of the text, observing the information from different dimensions is possible.

By selecting a keyword, an emotion, or a keyword, one can see the most and least common opinions and emotive reactions on them. The example below shows a dataset and grouping of the data according to predetermined aspects (De Bruyne et al., 2022).

We decided to work with aspect based emotion analysis, because it provides a detailed look at every emotion separately. As different emotions affect emotional connection differently, it is crucial to dive into. For more details on emotions and emotional connection, wait for our Emotional Connection Feature Blog Post that is coming soon

Aspects

1. Product - general 1,650

2. Social media 401

3. Product - availability 225

4. Product - variety 117

5. Company 94

6. Product - quality 35

7. Product - price 32

8. Website 29

9. Marcom 20

10. Personnel 12

• Total 2,615

An example from out product CogniScope below, shows the Aspect-based Emotion Analysis results, the relations between aspects and emotions with percentages can be observed.

Conclusion

In conclusion, we're thrilled to bring you the power of aspect-based emotion analysis. With this tool, we can explore text like never before, diving into emotions and opinions. Our mission is to help you understand your customers better, to uncover what makes them tick and where improvements are needed. Together, we can use this invaluable insight to build stronger connections and enhance our products and services. Thank you for joining us on this journey towards a deeper understanding of customer emotions and behavior.

References

Alqaryouti, O., Siyam, N., Monem, A. A., & Shaalan, K. (2019). Aspect-based sentiment analysis using smart government review data. Applied Computing and Informatics. Aspect-based sentiment analysis using smart government review data

Chuisano, F. (August 19, 2022). Quick intro to aspect-based sentiment analysis. Medium. Quick intro to Aspect-Based Sentiment Analysis

Cortal, G., Finkel, A., Paroubek, P., & Ye, L. (2022). Natural language processing for cognitive analysis of emotions. Semantics, Memory and Emotion. Natural Language Processing for Cognitive Analysis of Emotions

De, S., Dey, S., Bhatia, S., & Bhattacharyya, S. (2022). Advanced data mining tools and methods for social computing. Academic Press. An introduction to data mining in social networks

De Bruyne, L., Karimi, A., De Clercq, O., Prati, A., & Hoste, V. (2022). Aspect-based emotion analysis and multimodal coreference: a case study of customer comments on Adidas Instagram posts. Proceedings of the 13th Conference on Language Resources and Evaluation (LREC 2022), pages 574–580.