Artículos disponibles en nuestro repositorio digital

Data-Driven Usability Refactoring: Tools and Challenges

Usability has long been recognized as an important software quality attribute and it has become essential in web application development and maintenance. However, it is still hard to integrate usability evaluation and improvement practices in the software development process. Moreover, these practices are usually unaffordable for small to medium-sized companies. In this position paper we propose an approach and tools to allow the crowd of web users participate in the process of usability evaluation and repair. Since we use the refactoring technique for usability improvement, we introduce the notion of “data-driven refactoring”: use data from the mass of users to learn about refactoring opportunities, plus also about refactoring effectiveness. This creates an improvement cycle where some refactorings may be discarded while others introduced, depending on their evaluated success. The paper also discusses some of the challenges that we foresee ahead.

Citar como: Alejandra Garrido, Sergio Firmenich, Julián Grigera, Gustavo Rossi: “Data-Driven Usability Refactoring: Tools and Challenges”. 6th International Workshop on Software Mining (SoftwareMining 2017), co-located with ASE 2017. Urbana, Illinois, USA. Oct. 30, 2017. pp. 52-55. IEEE.

El artículo puede descargarse siguiendo este link

Kobold: web usability as a service

While Web applications have become pervasive in today’s business, social interaction and information exchange, their usability is often deficient, even being a key factor for a website success. Usability problems repeat across websites, and many of them have been catalogued, but usability evaluation and repair still remains expensive. There are efforts from both the academy and industry to automate usability testing or to provide automatic statistics, but they rarely offer concrete solutions. These solutions appear as guidelines or patterns that developers can follow manually. This paper presents Kobold, a tool that detects usability problems from real user interaction (UI) events and repairs them automatically when possible, at least suggesting concrete solutions. By using the refactoring technique and its associated concept of bad smell, Kobold mines UI events to detect usability smells and applies usability refactorings on the client to correct them. The purpose of Kobold is to deliver usability advice and solutions as a service (SaaS) for developers, allowing them to respond to feedback of the real use of their applications and improve usability incrementally, even when there are no usability experts on the team. Kobold is available at: A screencast is available at

Citar como: Julián Grigera, Alejandra Garrido and Gustavo Rossi, «Kobold: web usability as a service», in Proceedings of the 32nd {IEEE/ACM} International Conference on Automated Software Engineering, {ASE} 2017, Urbana, IL, USA, October 30 – November 03, 2017, , 2017, , Eds., . pp. 990–995

El artículo puede descargarse siguiendo este link

The use of glossaries to define consistent models on agricultural contexts

The agricultural practices vary in the different regions of the world. The diversity of weather conditions, soil, flora, and fauna, demands, for example, specific irrigation system and plant health check. Moreover, different countries have different regulations about the use of material (i.e., plastic and wooden), chemicals and techniques. Thus, even though the harvest cycle is the same: soil and material preparation, fertilization, disease management, and so on; every step is performed differently. In order to perform global analysis and find solutions to be applied in several contexts simultaneously, it is necessary to relate and find the equivalences among the different practices. This work proposes a technique to analyze descriptions in natural language of the different agricultural practices with the aim of finding similarities and equivalences. Thus, a unified description can be used as input to produce mathematical models. In particular, we use the Language Extended Lexicon (LEL), a glossary that defines symbols (words and phrases), through two attributes: notion and behavioral responses. The notion provides a description as a regular dictionary does, while the behavioral responses describe the tasks related to the symbol. The proposed technique consists in analyzing the similar patterns in the behavioral responses to find the equivalences. We have worked with only one language, but we think that we can extend our work to be used in a multilanguage environment.

Citar como: Leandro Antonelli, Alejandro Fernández, Jorge E. Hernández, “The use of glossaries to define consistent models on agricultural contexts”, Euro2018, 29th European conference on operational research, Valencia 8-11 julio 2018.

El artículo puede descargarse siguiendo este link

Gestalt Prototyping Framework – Evaluation Tool

The Gestalt Prototyping Framework linked the Nielsen Usability Heuristics to the principles of human perception and presents some parameters that can be used in developing high and low-fidelity prototypes for mobile application interfaces. The link between the fundamentals of usability and Gestalt principles focuses on the graphical components of the interfaces and the functions that they fulfill in the development of different functional actions. Previous articles have presented promising experimental results in reducing trial-error regressions in the interface design process and improving interface redesign processes; There have been positive results in parameters such as learnability, ease of use, perception of simplicity, and user preference. Within this same line of research, the present work describes the development process of an application that summarizes the fundamental aspects of the Gestalt prototyping framework in an evaluative model, which helps application designers, software engineers, and usability experts, to assess the prototypes of the interfaces and the incidence of the different graphic components in usability interactions. The tool is based on the most widely used usability parameters and is structured based on the responses that the development team fills in an automated LIKERT assessment. The results are processed based on an algorithm that simulates and predicts the results that could be obtained in high-fidelity prototype tests. This application issues specific recommendations on the visual components of the interfaces, to obtain better results in the production of high and low-fidelity prototypes.

Citar como: Daniel Ripalda, César Guevara and Alejandra Garrido: Gestalt Prototyping Framework – Evaluation Tool. In: Advances in Intelligent Systems and Computing. (2021), 747–752

El artículo puede descargarse siguiendo este link