Banca de QUALIFICAÇÃO: JOÃO BATISTA DE SOUZA NETO

Uma banca de QUALIFICAÇÃO de DOUTORADO foi cadastrada pelo programa.
DISCENTE : JOÃO BATISTA DE SOUZA NETO
DATA : 24/05/2019
HORA: 10:00
LOCAL: Auditorio I do DIMAp e https://www.appear.in/mmusicante
TÍTULO:

Mutation Test for Big Data programs

PALAVRAS-CHAVES:

Big Data; Mutation Test; Apache Spark; Taxonomy; Mutation Operators.

PÁGINAS: 140
GRANDE ÁREA: Ciências Exatas e da Terra
ÁREA: Ciência da Computação
SUBÁREA: Metodologia e Técnicas da Computação
ESPECIALIDADE: Linguagens de Programação
RESUMO:

The growth in the volume of data generated by society, its continuous and large-scale production and its heterogeneity led to the development of the Big Data concept.
Increasingly open access to these data presents challenges in the collection, storage and, above all, in its processing and analysis, which require important computing resources and adapted execution conditions.
Different parallel programming models and systems have been proposed for the processing of Big Data, such as the MapReduce model and its implementation in the Hadoop system, and the Dryad, Nephele, and Apache Spark systems, among others.
These systems, designed for cluster-type architectures, provide parallel scheduling and execution environments that hide associated technical difficulties (such as fault tolerance and data distribution, for example) and allow for a focus on the algorithmic aspects of data processing.
Regardless of the model and system adopted, Big Data processing applications need to be tested and evaluated, especially taking into account the costs involved in execution in that context.
However, the testing area at Big Data is still new, with few jobs that seek to apply systematic testing techniques.
This proposed doctoral thesis aims to reduce the gap in the testing area of Big Data processing applications by proposing a Mutation Test approach.
This is a test technique that seeks to simulate defects in a program by inserting modifications in its code, by applying the so-called mutation operators that determine how the modification is done, in order to create different versions of this one, called mutants.
These mutants can then be used both to evaluate a set of tests, to check how many defects this set can identify, and to design tests, in order to create tests that can identify the simulated defects.
This paper proposes a mutation test approach based on a model that covers the main characteristics of Big Data processing systems and defects that may appear in this context.
To identify the types of defects that may appear in Big Data processing programs, an investigation was made into defects and problems related to the Apache Spark system.
This study resulted in the development of two taxonomies.
The first taxonomy groups and characterizes nonfunctional problems that affect the execution performance of a Spark application.
The second taxonomy is focused on functional defects that affect the behavior of Spark applications.
A definition of mutation operators and a strategy for the generation of mutants are derived from this second taxonomy, allowing to design an approach for mutation testing of Big Data processing applications.

MEMBROS DA BANCA:
Presidente - 1221251 - MARTIN ALEJANDRO MUSICANTE
Interno - 1495704 - UMBERTO SOUZA DA COSTA
Externa à Instituição - ANAMARIA MARTINS MOREIRA - UFRJ
Externa à Instituição - GENOVEVA VARGAS-SOLAR - CNRS
Externa à Instituição - SILVIA REGINA VERGÍLIO - UFPR