Search

This work shows the learning results obtained from the use of a technique proposed and applied by the authors called Literature Based Modeling Learning. This technique consists of the development of modeling tasks where the input is a text that represents a literature piece and the output is a set of diagrams that model that LP. The literature piece is typically a well-known short story but it can also be the lyrics of a song or hymn or even a prayer. This technique has been used during the last eight semesters at a Mexican university. A results of a study are presented to identify possible benefits with the use of this technique.

Source code metrics have been used as indicators of code comprehension according to its complexity. Shao and Wang proposed a cognitive complexity measure based on the cognitive weight of the basic control structures. This cognitive weight is a fixed value assigned to each control structure, which was calculated by means of a series of empirical experiments. Given the lack of information and incompleteness of the experiments, improvements to these experiments must be made before a measure can be used. In this paper, a study on the overall state of the cognitive weight, including its experiments, is presented, along with an empirical experiment which will help us to assess the validity and accuracy of the cognitive weight. The experiment consists of applying a survey to 114 students. It contains source code complexity and comprehension questions that allow us to obtain conclusions about the validity of the cognitive weight.

Computer programming is one of the basic abilities developed in computing science students, but it is important to note that programs are the product of generating code from software design. This work presents the results of examining the design products of students in different stages of their degree program. We have replicated some experiments reported in the literature but a difference between these works and ours is that we analyze the design work of students at all levels of their degree programs. We find results that indicate that the design abilities do not necessarily show a continuous improvement during the passage of time in their studies and we explore possible causes of this and outline future research in this line of investigation.

Software quality measurement is an essential task in any software related project. Throughout the years many software quality models have been proposed defining a set of quality elements that have to be measured in order to evaluate the software quality. However currently, there is no standard to measure any of the non-functional requirements, instead there are models that provide a characterization for a particular non-functional requirement, including the software quality (ISOnEC 25000). Even if this model provides a common base for quality measurement, most of the defined characteristics are qualitative, thus their measurement are complicated, very often subjective, with a limited scope mostly particular, and not necessarily follow the measurement principles, which limiting the comparison. Additionally, uncertainty and lack of information are also present in software projects, making it harder to measure the overall software quality. Because of these issues, better methods for software quality evaluation have to be used. In this paper a fuzzy logic-based model is proposed for quality evaluation, aiming to handle in a formal way the uncertainty, the lack of information and enable the possibility to compare quantitatively the quality evaluation at distinct levels. The model was tested by measuring nine real life projects and obtaining conclusions about the quality of each project.

Source code metrics extraction is a complex task that has to be done automatically given the current size of software. They are extracted using software metric tools and more generic extraction mechanisms. These mechanisms usually work by querying a source code representation model. These models are static, and the information that can be obtained from them is limited. In this work an extraction methodology is presented in which the model is created every time certain information is needed. This is accomplished by querying the language context-free grammar, and from the information obtained by the query, a dynamic model is created. Current extraction mechanisms work by querying a model, while the proposed methodology queries the grammar directly, thus the model is created afterwards from the query result, and contains all the needed information. A metrics tool is created based on the proposed methodology, and in order to prove the correct functioning of extracting the desired information from the source code, not as already predefined as in current tools, several metrics are extracted as defined by four existing metrics tools. Querying the language grammar allows access to all available data in the source code, regardless of the programming language and paradigm.

Current research on source code metrics is heavily focused on measuring quality attributes for object oriented source code, for common languages such as $C$ ++, Java and C#. However, source code metrics are good predictors and evaluators of software systems characteristics, thereby researchers have found other uses and applications for other computing related areas. In this research, source code metrics for Very High Speed Integrated Circuit Hardware Description Language (VHDL) are proposed and used to predict the synthesized product properties for Field Programmable Gate Array (FPGA) based digital systems. Code written in VHDL is used to generate a configuration file for a specific FPGA in a process named design synthesis. Physical properties of the FPG/VHDL based synthesized product are measured for performance evaluation in a testing process, but it can be a time consuming process. In this paper we aim to correlate source code metrics with the FPGA/VHDL based synthesized product properties, in order to determine if source code metrics can be used as predictors of certain synthesized product properties. A case of study correlating three source code metrics with three synthesized product properties is presented. The results of the study provide evidence that source code metrics can be used as predictors of FPGA/VHDL based synthesized product properties.

Software development is a human activity typically performed by individuals or groups working in an organization. For this task to be successful, a variety of human factors must be considered, from individual factors (cognitive capacity or motivation of the developer) to group factors (ease of teamwork or to participate in social interaction) This research focuses on the study of the possible correlation between the personality of the software engineer and their individual specific capacity for software design. The fundamentals of characterization of currently accepted personality features and a comparison of these features in the adult population in Mexico, Canada, United States and Brazil are presented. Finally, the results of an experiment with computer engineering students from a Mexican university are presented, relating these characteristics with their ability to develop software design products.

Software requirements are essential for the correct development and planning of a software project. Each requirement is related to a software quality category, i.e. usability or maintainability, and their classification into these categories could greatly help the requirements analysis process. Requirements are usually expressed in natural language as written documents and many methods have been proposed for their automatic analysis and classification, based mainly on word frequency analysis. In this paper, a method for extracting keywords from users' written requirements using the TextRank technique and inverse frequency analysis is presented. These keywords represent relevant computing-related terms that can be mapped to a certain quality category which allows us to identify core terms that are of major relevance in the text of a given requirement. A total of 946 software requirements from six online datasets were analyzed and 390 keywords were extracted. The quality categories defined in the ISO/IEC 25000 standard will be used for keyword classification.

During the software development cycle, software products can undergo many changes in requirements and design. In one of those products, the source code of the system, these changes can impact the original specifications and design, leading to tangled and scattered code that affects the design and maintenance processes. Research efforts have been made to analyze the scattered code for concerns identification and extraction. In object oriented systems, these concerns can be distributed across many classes, leading to the definition and identification of crosscutting concerns. Many methods have been proposed to identify, extract and separate crosscutting concerns. In this paper, a method for crosscutting concerns identification using information retrieval techniques is presented. This work aims to identify the core crosscutting concerns, which are the main tasks a system has to satisfy whose code is scattered over the different classes of the system. Information retrieval provides advantages over common parsing techniques since the code is treated as unstructured text, making the method programming language independent.

Object tracking refers to the relocation of specific objects in consecutive frames of a video sequence. Presently, this visual task is still considered an open research issue, and the computer science community attempted solutions from the standpoint of methodologies, algorithms, criteria, benchmarks, and so on. This article introduces a GPU-parallelized swarm algorithm, called the Honeybee Search Algorithm (HSA), which is a hybrid algorithm combining swarm intelligence and evolutionary algorithm principles, and was previously designed for three-dimensional reconstruction. This heuristic inspired by the search for food of honeybees, and here adapted to the problem of object tracking using GPU parallel computing, is extended from the original proposal of HSA towards video processing. In this work, the normalized cross-correlation (NCC) criteria is used as the fitness function. Experiments using 314 video sequences of the ALOV benchmark provides evidence about the quality regarding tracking accuracy and processing time. Also, according to these experiments, the proposed methodology is robust to high levels of Gaussian noise added to the image frames, and this confirms that the accuracy of the original NCC is preserved with the advantage of acceleration, offering the possibility of accelerating latest trackers using this methodology.

The development of quality software is a basic requirement that must be observed. Measuring software is a tool that allows the development of quality software for its entire life cycle. For software measurement, software metrics are used, among other techniques, which allow us to obtain a numerical value from a software product. There are two problems with these measurements: a value obtained can have different meanings depending on the project and what is desired as a result from the measurement, and the other problem is that the number and type of measurements is limited by the capabilities of the used tool. This paper presents a promising solution to the problem above by presenting a technique with which users can obtain any desired metrics and apply them to code in any programming language.

Software metrics are an essential tool in the software measurement process. As software systems grow, and in consequence their corresponding source code, automated tools for source code metrics extraction, usually known as Software Metric Tools, should be developed in order to aid the user to correctly achieve the process. Even though these tools are of great help, they have certain issues and are limited by the set of accepted metrics and programming languages. This paper propose a new type of metric tool aimed to be user oriented, meaning that the tool provides mechanisms that allow the user to define his own metrics and incorporate new languages. A tool with these characteristics requires more complex methodologies for its construction. In this paper we describe the problematic of current software tools and the construction of a user oriented tool which presents advantages over the current tools.

La calidad de los programas de software (SW) es considerada cuando éstos se realizan bajo estrictas metodologías y siguiendo estándares que intentan reducir la probabilidad de presentar defectos. La obtención de métricas de SW es un método útil para minimizar dicha probabilidad. Para evaluar la calidad de un programa de SW, uno de los indicadores más importantes es la mantenibilidad, la cual se puede medir mediante la obtención de un conjunto de métricas. La mantenibilidad es la facilidad con la que un sistema de SW puede ser modificado y es un atributo que afecta de manera crucial al costo de desarrollo del mismo. La literatura muestra estudios sobre mantenibilidad y la alteración de este parámetro cuando los programas evolucionan. Este trabajo muestra los resultados de analizar 315 programas realizados por estudiantes universitarios de carreras en Ingeniería en Computación e Informática y la evolución que la mantenibilidad y otras métricas básicas presentan en función del progreso en la formación académica de sus autores.