# A dataset of interactions and emotions for website user experience evaluation

## TL;DR

This paper introduces a dataset combining website interaction logs with emotional data to study how emotions affect user behavior and improve AI-driven user experience.

## Contribution

The novel contribution is a dataset linking user interaction data with Ekman-encoded emotional responses for affective computing applications.

## Key findings

- The dataset includes 30 days of interaction logs paired with emotional data using Ekman’s model.
- It enables analysis of how emotions influence user behavior across different websites.
- The dataset supports developing AI systems that detect emotions from interaction data alone.

## Abstract

This data descriptor introduces a dataset designed for affective computing applications in the context of human-computer interaction, with a particular focus on user experience (UX) and website interactions. The dataset comprises interaction logs, including mouse movements and key presses, collected over a period of 30 days from various websites. Each recorded interaction is paired with corresponding emotional data, which is encoded according to Ekman’s emotion model, allowing for a nuanced analysis of emotional responses. This dataset is particularly valuable for examining how emotions influence user behaviour, and vice versa, also across different types of websites. Its primary aim is to facilitate the development of Artificial Intelligence (AI) systems capable of detecting user emotions solely based on user interaction data. Such systems have potential applications in improving UX design, personalizing web content, and enhancing the overall UX by adapting to emotional states without requiring explicit input from users.