This paper presents the design and implementation of a system for semi-automating the literature review process based on machine learning. By using machine learning algorithms, the system determines whether scientific papers belong to the topic that is being explored as part of the review process. The system’s user interface allows the process of creating a literature review to be managed through a series of steps: selecting data sources, building queries and topic searches, displaying the scientific papers found, selecting papers that belong to the set of desired papers, running machine learning algorithms for learning and automated classification, and displaying and exporting the final set of papers. Manual literature reviews are compared with automated reviews, and similarities and differences between the two approaches in terms of duration, accuracy, and ease of use are discussed. This study concludes that the best results in terms of sensitivity and accuracy for the automated literature review process are achieved by using a combined machine learning model, which uses multiple unweighted machine learning models. Cross-testing the models on two alternative datasets revealed an overlap in the machine learning hyperparameters. The stable sensitivity and accuracy in the tests indicate the potential for generalized use in automated literature review.
10 December 2018 in Journal of information and organizational sciences
Proactive-reactive scheduling is important in the situations where the project collaborators need to coordinate their efforts. The coordination is mostly achieved through the combination of the shared baseline schedule and the deviation penalties. In this paper, we present an extension of predictive Gantt chart to the proactive-reactive scheduling needs. It can be used to track the evolution of the relationship between dynamic and static elements through the time. The dynamic elements are evolving probability distributions due to the uncertainty and revealed information. The static elements are time-agreements in the baseline schedule. We demonstrate that in the state-of-the-art proactive-reactive scheduling, the baseline schedule is agnostic to the information received during the project execution. The sources of such inflexibility in the problem model and the scheduling methods are analyzed. The visualization is highlighted as a precursor to developing new methods that proactively change the baseline schedule in accordance with the gained information.
Dynamic software updating (DSU) is an extremely useful feature to be used during software evolution. It can be used to reduce down-time costs, for security enhancements, profiling and testing new functionalities. There are many studies and solutions on dynamic software updating regarding diverse problems introduced by the topic, but there is a lack of research which compares various approaches concerning supported changes and demands on resources. In this paper, we are comparing currently available concepts for Java programming language that deal with dynamically applied changes and measuring the impact of those changes on computer resource demands.
Tree comparisons are used in various areas with various statistical or dissimilarity measures. Given that data in various domains are diverse, and a particular comparison approach could be more appropriate for specific applications, there is a need to evaluate different comparison approaches. As gathering real data is often an extensive task, using generated trees provides a faster evaluation of the proposed solutions. This paper presents three algorithms for generating random trees: parametrized by tree size, shape based on the node distribution and the amount of difference between generated trees. The motivation for the algorithms came from unordered trees that are created from class hierarchies in object-oriented programs. The presented algorithms are evaluated by statistical and dissimilarity measures to observe stability, behavior, and impact on node distribution. The results in the case of dissimilarity measures evaluation show that the algorithms are suitable for tree comparison.