ABSTRACT Software development effort estimation (SDEE) is essential for effective project planning and relies heavily on data quality affected by incomplete datasets. Missing data (MD) are a prevalent problem in machine learning, yet many models treat it arbitrarily despite its significance. Inadequate handling of MD may introduce bias into the induced knowledge. It can be challenging to choose optimal imputation approaches for software development projects. This article presents a novel incomplete value imputation model (NIVIM) that uses a variational autoencoder (VAE) for imputation and synthetic data. By combining contextual and resemblance components, our approach creates an SDEE dataset and improves the data quality using contextual imputation. The key feature of the proposed model is its applicability to a wide variety of datasets as a preprocessing unit. Comparative evaluations demonstrate that NIVIM outperforms existing models such as VAE, generative adversarial imputation network (GAIN), ‐nearest neighbor (K‐NN), and multivariate imputation by chained equations (MICE). Our proposed model NIVIM produces statistically substantial improvements on six benchmark datasets, that is, ISBSG, Albrecht, COCOMO81, Desharnais, NASA, and UCP, with an average improvement in RMSE of 11.05% to 17.72% and MAE of 9.62% to 21.96%.

Time-related properties are a critical type of extrafunctional requirements for designing real-time systems. Modeling and validating time-related properties at the requirements specification and analysis phases is important for the successful development of real-time systems in terms of cost, quality and productivity. In the literature and practice, timing analyses (e.g., Worst Case Execution Time) are often performed to ensure that the design of a real-time system fully conforms to its time-related constraints. However, such analyses are mostly performed at the design and implementation stages, but not at the requirements level. This paper presents a restricted, natural language based, use case modeling methodology (named as RUCM4RT) to specify functional requirements of real-time systems as use case models, along with associated time-related constraints. RUCM4RT was proposed based on the UML profile for Modeling and Analysis of Real-Time and Embedded Systems (MARTE). In addition, in this paper, we also propose a metamodel-based formalization mechanism named as UCMeta4RT to automatically formalize use case models. We have conducted two real-world case studies to evaluate our solution and 40 use cases were modeled, among which 27 realtime use cases, 118 time-related constraints and 47 other extrafunctional (also commonly called non-functional) constraints were specified. Results show that RUCM4RT was able to handle all the real-time related elements (e.g., time-related constraints) of the use case models.

To reduce the cost and risk of the development of system of systems (SoS) by pre-evaluation before the SoS is built, a model-driven approach is proposed to evaluate the SoS based on its architecture, especially focused on measures of performance and effectiveness. In order to implement the pre-evaluation, the system architecture needs to be transformed to the simulation model of the system under the directing of the evaluation requirement model, then the evaluation can be done based on the evaluation model and the data from simulation. The architecture, evaluation requirement model, simulation model and evaluation model involved in the above process constitute the model system (a set of models related with each other) of our approach. Through the modeling activities based on the model system, we can implement the systematically pre-evaluation of SoS. The model system and its meta-model, which are the core of our approach, are detailed after the introduction of the framework of our approach. Then the transformations from DoDAF architecture to evaluation requirement model and simulation model are studied for accelerating the evaluation process. The case study shows that our approach can not only ensure the standardization and systematization of the evaluating process, but can improve the evaluating efficiency and creditability based on the proposed transformation method.

In recent years, collaborative filtering becomes one of the most successful recommender systems. Its key technique is to predict new ratings from the known ratings. Unfortunately, in the previous research, the temporal information was rarely applied. That is to say, the ratings at different time were considered the same. However, from our point of view, not only the mean values of ratings in different periods are different, but users' opinions toward items may change with the passage of time as well. We analyze the influence of the temporal information and introduce three methods to apply the temporal information. Firstly, by mixing user, item and time attributes, we present a regression-based method. Secondly, to guarantee that the ratings in different time can contribute different weights to the predicting rating, we adjust the prediction function by adding a parameter, which is a function of the time between the predicting rating and the known rating. Thirdly, we select different methods to predict ratings of different periods. Experiments on two real large datasets show that our methods are effective and can improve the accuracy.

The invalidated input is one of the most critical web application security flaws. However, testing the user input validation function is an intellectual and labor intensive task. We are developing a model driven framework to help testers to accomplish this job in visual view with guidance. This paper reports our on-going work. A meta-model of Web application for user input validation testing is defined. Based on the meta-model, by analyzing HTML files, a light weight method is given to create the model. Our evaluation shows that the proposed method can comprehensively model Web applications, and accurately identify the purpose of input points, which are very important for the test case generation in the future.

Web applications have become complex and crucial in many fields. In order to assure their quality, a high demand for systematic methodologies of Web application testing is emerging. In this paper, a methodology of model-driven testing (MDT) for Web application is presented. Model is the core of this method. Web application model is built to describe the system under testing. Test case models are generated based-on the Web application model. Test deployment model and test control model are designed to describe the environment and process of test execution. The test engine executes test cases based-on the test deployment and control model automatically. After that, testing results are reflected to test case models. A framework is designed for supporting this methodology. In order to get better extensibility and flexibility, it is loose-coupled by a modeler and a tester. The modeler enables developers to design meta-models, and is responsible for creating, visualizing and saving models. The tester takes in charge of recovering the tested Web application model, generating test cases, and executing test

This paper discusses the formalization of the test adequacy criterion for context-free grammars. The basic concept of grammar-based testing is introduced and the grammar-based test criterion is formally defined. This paper also proposes a family of grammar-based structural test criteria whose subsume relations and test set sizes are analyzed, which are used to assess the proposed criteria. Some basic properties based on Weyuker's axiom systems are generalized and some fundamental properties specific to grammar-based testing are proposed We believe the concepts developed in this paper will contribute to the classical theory of software testing adequacy and are very valuable for software test practice

Predicting failures from software input is still a tough issue. Two models, namely the surface model and structure model, are presented in this paper to predict failure by applying the maximum entropy principle. The surface model forecasts a failure from the statistical co-occurrence between input and failure, while the structure model does from the statistical cause-effect between fault and failure. To evaluate the models, precision is applied and 17 testing experiments are conducted on 5 programs. Based on the experiments, the surface model and structure model get an average precision of 0.876 and 0.858, respectively