Sistema Regional de Información
en línea para Revistas Científicas de América Latina,
el Caribe, España y Portugal

Objective: Requirements validation is an important software development activity since it allows to business and technical stakeholders agree on functional and non-functional software requirements in order to these being complete, consistent and unambiguous. An appropriate non-functional requirements validation supports to software product quality because this kind of requirements are considered as aspects that contribute to this quality. In this sense, this article addresses difficulties that arise during the software development for an inadequate non-functional requirements validation by means on knowledge management approach. Methodology: A conceptual model and a process for non-functional requirements validation are proposed using the action-research methodology multicycle with bifurcation.  Results: Through the focus group method that involved professionals with experience in requirements validation this proposal has been preliminarily evaluated and positive results regarding the relevance, applicability and clarity of the validation process has been obtained. The contribution of validation techniques in activities as development planning and software product testing is highlighted, since they support the software development process productivity. Conclusions: The validation process integrates specific elements of knowledge management allowing the simplification of the identification, capture, specification and even non-functional requirements analysis, by establishing a structured unique format which can be managed during the execution of these practices.

Objective: To discover patterns related with the performance of high school students who presented the Critical Reading test on Saber 11 exam between 2015 and 2016, using the classification technique based on decision trees. Methodology: The CRISP-DM was used as methodology. From the data available in the ICFES ftp services about the results of the Saber 11º tests, a data set was built. This data set was preprocessed and transformed. The result obtained by the students in the Critical Reading test was selected as target attribute. A classification model based on decision trees was obtained, using the Weka data mining tool. Results: Among the factors related with performance in Critical Reading and that are an integral part of the patterns discovered are: the age, the stratum, study day and ICT conditions of the student. Conclusions: A good academic performance on the critical reading test is associated with students from high socioeconomic strata, students who attend full-time or single-day schools, as well as students with good ICT conditions. This means that, to improve the quality of basic and secondary education in Colombia, it is necessary for the national government, through the MEN, to implement the single day in most public schools, with a necessary endowment that allows students from the lower strata find better study conditions than the upper strata, thus reducing the social gap that currently exists.

Objective: To propose a CLR (Clusterwise Linear Regression) algorithm that carries out a divisive grouping of pavement segment samples using linear regression models and automatically defines the number of groupings in order to predict the pavement serviceability index (PSI). Methodology: Based on the iterative research process proposed by Pratt, two improvement cycles of the proposed algorithm were developed. The first cycle allowed us to obtain an initial version, apply it to the training and test datasets, and observe the improvements that had to be made. In the second cycle, the final version was obtained to which the parameters were tuned, and it was compared with the state of the art using several metrics. Results: A model composed of three groups of pavement segment samples was obtained with their corresponding multivariate linear regression models (mixed attributes) that allow predicting the PSI of a pavement sample. Conclusions: The model was obtained with less computing time (15.6 times less time than that reported by the state of the art) and presents better results in simplicity compared to the linear and non-linear models reported in the literature, in addition, it has quality results similar (even better in some metrics) to the linear model and is competitive against the non-linear model.

Objective: Propose an IoT architecture for obtaining, monitoring and taking advantage of data pertaining to physiological variables of a patient (heart rhythm and oxygen level) through commercial wearable devices for health care service.Methodology: Taking as a reference the conventional four-layer IoT architecture (capture, storage, analysis and visualization), a three-view architecture (business, functional and implementation) was designed for the development of physiological variables monitoring services that take advantage of the data from the commercial devices for obtaining value-added information aimed at specialized personnel.Results: As a result, a reference architecture was obtained, which was validated by developing an IoT system for monitoring and analyzing of both the blood oxygen level and heart rate.Conclusions: The results obtained demonstrate that the proposed architecture serves as a reference for the construction of monitoring and analysis systems of physiological variables in the health assistance service.

Objective: Present a reference model that facilitates the understanding and implementation of Scrum in software organizations. Methodology or method: The design of the model was based on the comparison and identification of the fundamental elements of the existing agile guides most used in the software industry. Likewise, the GQM (Goal-Question-Metric) paradigm was used, which allowed detailing the reference model through a clear and complete set of process elements, as well as: phases, activities, roles, expected / satisfactory results and flows process in BPMN. Using the GQM paradigm allows the model to be extended in the future with the design of an assessment / evaluation instrument that allows measuring the degree of Scrum compliance. The evaluation of the model was carried out through a focus group, where the relevance, clarity and completeness of the proposal was evaluated, the focus group was made up of participants who are experts in Scrum. Results: It was evident that the participants consider that the proposed model is complete, suitable and clear, and that it would facilitate the understanding, adoption and evaluation of Scrum. Conclusions: In short, a model is proposed that suggests six (6) phases, twenty-nine (29) activities, five (5) roles and other process elements that would allow professionals and organizations to be guided to minimize subjectivity in understanding and adopting Scrum. As future work, it is expected to define an evaluation instrument based on the reference model presented here, which allows evaluating the degree of Scrum compliance.

Objective: Train Neural Networks and Support Vector Machines for regression for predicting future trends in kidnapping victims numbers by years in Colombia, with the purpose of support decision making. Methodology: A dataset about kidnapping victims with 39.863 records were processed, the dataset was taken from the government open data given by the Colombian Prosecutor's Office, with means of training and evaluating a Neural Network and a Support Vector Machine for regression (SVR), this allowed to identify the model with less Root Mean Squared Error (RMSE) for predicting kidnapping victims trends. Results: We found that the trained Neural Network had less Root Mean Squared Error (RMSE) against the Support Vector Machine and we were able to estimate the number of kidnapping victims in the next ten years throughout the Colombian territory. Conclusiones: There are several papers related to crime prediction, however, this study proposes a new model of Neural Networks to estimate the trend of kidnapping victims in Colombia. The model allows us to predict that over time, the number of kidnappings will decrease.

Objective: Present a state of the art around problems that arise when creating complex interaction services between sensors on the web, such as: automatic discovery, combination of heterogeneous data, coordination and cooperation between sensors, solutions with the development of semantic middlewares for the web of Things, without achieving transparent cooperation between them, deepening contributions in machine-to-machine interaction. Methodology: The research was of a documentary type through a bibliographic analysis of the last 12 years, identifying thematic nuclei, selection criteria, descriptive sheets, abstracting them for their development. Results and conclusions: The results exhibited by thematic core the main components of the topic, problems a discussion around the challenges by resolution in this research area, analyzing current approaches in the current state of knowledge related to the semantic interaction of the Web of things (WoT) .

Objective: To propose a data warehouse with high capacity for analysis for the performance of university students; analyzing the different design situations that may be common for higher education institutions, allowing to obtain more effective models and facilitating their adaptation. Methodology or method: Common dimensions and measures found in the literature on data warehouses related to students' academic performance were identified, then elements typical of the country's universities were included. Results: Two dimensional models for the academic performance of students in higher education institutions and the common situations analyzed in the design process of these. In addition, a prototype of the data warehouse of these models for the University of Cauca and an evaluation based on the ISO / IEC 25022 standard, of the level of user satisfaction with the models implemented. Conclusions: The results of the evaluation showed that 87.72% of the users were satisfied with the data provided by the model through the prototype, observing the importance of this type of system in universities.