| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 3.17 MB | Adobe PDF |
Authors
Abstract(s)
O recente aumento da oferta de aplicações de software contribuiu para o aumento da exigência
dos utilizadores, o que originou um acréscimo da complexidade dessas aplicações. Tal
contribuiu para a adoção de mecanismos de automatização do processo de testes de software,
de forma a reduzir erros de codificação e encurtar o tempo necessário para disponibilizar uma
nova versão da aplicação ao utilizador. Atualmente, automatizar o processo de testes de
aplicações é um processo já bem estudado e suportado por muitas ferramentas. No entanto, a
avaliação da usabilidade de uma aplicação requer soluções que permitam determinar,
antecipadamente, o tipo de melhorias que podem ser necessárias na aplicação sem a execução
de um elevado número de testes com utilizadores.
A performance é um fator tão importante numa aplicação web como a avaliação da sua
usabilidade. O número de interações cliente-servidor, bem como o volume de dados transferido
em cada interação estão relacionados com o desempenho da aplicação web, pelo que a
satisfação do utilizador está dependente destes fatores.
O presente documento aborda a automatização da análise do impacto das alterações do desenho
da aplicação no utilizador, assim como a análise de dependências entre interações clienteservidor. Através da implementação do modelo Keystroke Level Model (KLM) são medidas
interações não planeadas do utilizador, sendo posteriormente identificados padrões com base
nos operadores KLM e nos tempos de cada operador. Com base nestes padrões são gerados
indicadores de desempenho que são analisados pelo designer de interfaces, possibilitando
alterações no desenho da aplicação.
De forma a facilitar a análise da performance das aplicações web, é também proposta uma
ferramenta que identifica os recursos que podem ser transferidos do servidor com a
antecedência necessária de forma a evitar atrasos na interação do utilizador com a aplicação.
No sentido de priorizar o carregamento de recursos é identificada uma ordem para o seu
carregamento com base nas diferenças temporais entre eles.
ABSTRACT: The recent increase in the availability of software applications has contributed to the increase in user demand, which has led to an increase in the complexity of these applications. This contributed to the adoption of automation mechanisms for the software testing process, in order to reduce coding errors and shorten the time needed to make a new version of the application available to the user. Currently, automating the application testing process is a well-established reality and supported by many tools. However, the evaluation of the usability of an application requires solutions that allow to determine, in advance, the type of improvements that may be necessary in the application without the execution of a high number of tests with users. In a web application, its performance is a factor as important as the evaluation of its usability. The number of client-server interactions, as well as the volume of data transferred in each interaction are related to the performance of the web application, so user satisfaction is dependent on these factors. This work deals with the automation of the analysis of the impact on the user of changes in the design of the application, as well as the analysis of dependencies between client-server interactions. Through the implementation of the Keystroke Level Model (KLM), unplanned user interactions are measured, and patterns are subsequently identified based on the KLM operators and the times of each operator. Based on these standards, performance indicators are generated and analyzed by the interface designer, allowing changes in the design of the application. In order to facilitate the analysis of the performance of web applications, a tool is also proposed that identifies the resources that can be transferred from the server with the necessary advance in order to avoid delays in the user's interaction with the application. In order to prioritize the loading of resources, an order is defined based on the temporal differences between them.
ABSTRACT: The recent increase in the availability of software applications has contributed to the increase in user demand, which has led to an increase in the complexity of these applications. This contributed to the adoption of automation mechanisms for the software testing process, in order to reduce coding errors and shorten the time needed to make a new version of the application available to the user. Currently, automating the application testing process is a well-established reality and supported by many tools. However, the evaluation of the usability of an application requires solutions that allow to determine, in advance, the type of improvements that may be necessary in the application without the execution of a high number of tests with users. In a web application, its performance is a factor as important as the evaluation of its usability. The number of client-server interactions, as well as the volume of data transferred in each interaction are related to the performance of the web application, so user satisfaction is dependent on these factors. This work deals with the automation of the analysis of the impact on the user of changes in the design of the application, as well as the analysis of dependencies between client-server interactions. Through the implementation of the Keystroke Level Model (KLM), unplanned user interactions are measured, and patterns are subsequently identified based on the KLM operators and the times of each operator. Based on these standards, performance indicators are generated and analyzed by the interface designer, allowing changes in the design of the application. In order to facilitate the analysis of the performance of web applications, a tool is also proposed that identifies the resources that can be transferred from the server with the necessary advance in order to avoid delays in the user's interaction with the application. In order to prioritize the loading of resources, an order is defined based on the temporal differences between them.
Description
Keywords
Keystroke level model Interações Automatização Operadores Usabilidade Desempenho