A Framework for Identifying Sequences of Interactions That Cause Usability Problems in Collaborative Systems

Santos Bringas,Cristina Tîrnăucă,Rafael Duque,José Luis Montaña,Alicia Nieto-Reyes

doi:10.3390/electronics10040388

Santos Bringas, Cristina Tîrnăucă + Show 3 more

Open Access

https://doi.org/10.3390/electronics10040388

Copy DOI

Abstract

Collaborative systems support shared spaces, where groups of users exchange interactions. In order to ensure the usability of these systems, an intuitive interactions’ organization and that each user has awareness information to know the activity of others are necessary. Usability laboratories allow evaluators to verify these requirements. However, laboratory usability evaluations can be problematic for reproducing mobile and ubiquitous contexts, as they restrict the place and time in which the user interacts with the system. This paper presents a framework for building software support that it collects human–machine interactions in mobile and ubiquitous contexts and outputs an assessment of the system’s usability. This framework is constructed through learning that is based on neural networks, identifying sequences of interactions related to usability problems when users carry out collaborative activities. The paper includes a case study that puts the framework into action during the development process of a smartphone application that supports collaborative sport betting.

Highlights

Introduction and MotivationCurrent software methodologies promote user participation to evaluate the prototypes generated in each iteration of the life cycle development [1]
This framework is based on a three stage procedure: (i) define a model of the interactions supported by the system and how they are sequenced to carry out the user tasks; (ii) learn a neural network model that predicts usability metrics values using only information of the interactions that were performed by real users; and (iii) use the neural network model to identify the sequence of interactions that leads to usability problems, because of not being adapted to the individual or collaborative behaviour of the users
The usability evaluation methods are usually classified into four categories [6]: (i) heuristic methods—one or more experts inspect the system to find usability problems, (ii) observation methods—information related to users interacting with the system is collected, (iii) surveys—users answer questionnaires and/or interviews, and (iv) experimental methods—usability laboratories and software support, such as logging tools are used

Summary

Introduction

Current software methodologies promote user participation to evaluate the prototypes generated in each iteration of the life cycle development [1] This user feedback is a valuable source of information for eliciting and validating requirements. Laboratory usability evaluations do not reproduce these mobile and ubiquitous contexts, as the place and time in which the user interacts with the system is restricted. Harms and Grabowski [10] present a proposal that processes log files, generates task trees with the user actions, and identifies usability smells. The values that these usability metrics take must be interpreted in such a way that they have an impact on the set of iterations of the software development process. In the case of mobile and ubiquitous computing, these experimental methods must be applied in tests that reproduce the physical and social context in which the sys-

Results

Discussion

Conclusion