Análise experimental da eficácia de um método de inspeção da interface de usuário para predição da resposta emocional

Abstract

In the area of Human-Computer Interaction (IHC), evaluations help to know the desires and problems that people face when interacting with user interfaces (UI). Evaluations can be carried out in order to investigate aspects such as usability, accessibility, communicability of systems and the emotional response of users. There are some methods and ways of inspecting and classifying emotions evoked by UIs, such as physiological signals. However, most instruments for measuring physiological signals are expensive, invasive and use complex equipment. The Semiotic Inspection Method for evaluating Emotional responses (MISE) is a cheaper and less laborious alternative for evaluate the emotional state, as it does not require the participation of users to be applied. Several automatic searches were performed on search engines and few of the works found attempted to evaluate the effectiveness of an analytical method. For this reason opted for manual search, through the procedure of Snowballing. With this procedure, 32 studies were found that evaluated the effectiveness of analytical methods through comparisons with empirical methods. But, in most cases, the studies compared UIs usability methods and none analyzed methods of emotional response from users. There are studies in the literature that evaluate the feasibility of MISE, but so far, the effectiveness of MISE compared to other methods that evaluate the emotional response of users has not yet been measured. This research project aimed to evaluate whether MISE is effective to predict the emotional response evoked by UIs. For this, two empirical studies were carried out: (1) comparison of a gold standard (results from the application of MISE by the method proponent) with physiological signs and subjective feelings collected from 24 users and (2) comparison of the results of MISE applied by 9 IHC scholars versus a gold standard. In study 1, users had the signs of cardiac activity, brain activity and skin electro conductivity captured, while performing tasks in three types of UI: a game, a website and an application. In study 2, the researchers received training on the application of MISE and carried out inspections on the same 3 IUs used in the study with real UI users. At the end of the collection, comparisons were made between the empirical studies. The results suggest that the MISE proved to be easily apprehensible for HCI scholars, with no experience in applying the method. The data collected through physiological signals suggest that MISE can be effective in predicting the emotional state of users on site type UIs. And it is not effective on game and application type UIs. The markings of the volunteers in the collection of subjective feeling suggest that MISE can be effective in predicting the emotional state of users in game and application type UIs. And not effective on site type UIs.