Attributed graph clustering aims to group nodes into disjoint categories using deep learning to represent node embeddings and has shown promising performance across various applications. However, two main challenges hinder further performance improvement. Firstly, reliance on unsupervised methods impedes the learning of low-dimensional, clustering-specific features in the representation layer, thus impacting clustering performance. Secondly, the predominant use of separate approaches leads to suboptimal learned embeddings that are insufficient for subsequent clustering steps. To address these limitations, we propose a novel method called Semi-supervised Deep Attributed Clustering using Dual Autoencoder (SDAC-DA). This approach enables semi-supervised deep end-to-end clustering in attributed networks, promoting high structural cohesiveness and attribute homogeneity. SDAC-DA transforms the attribute network into a dual-view network, applies a semi-supervised autoencoder layering approach to each view, and integrates dimensionality reduction matrices by considering complementary views. The resulting representation layer contains high clustering-friendly embeddings, which are optimized through a unified end-to-end clustering process for effectively identifying clusters. Extensive experiments on both synthetic and real networks demonstrate the superiority of our proposed method over seven state-of-the-art approaches.

Recommender systems (RSs) are a significant subclass of the information filtering system. RSs seek to predict the rating or preference that a user would give to an item in various online application community fields. Collaborative filtering (CF) is a technique which predicts user distinctions by learning past user‐item relationships. However, it is hard to perceive the comparable interests between customers in light of the fact that the sparsity problem is caused by the deficient number of the relationship between users. It is a challenge which limited the ease of use of CF. This paper proposes a novel fuzzy C‐means clustering approach which is used to deal with this sparsity problem by utilising a sparsest sub‐graph detection algorithm in defining initial centres of the clustering method. The approach uses adaptability of fuzzy logic to make better personalised recommendations in terms of precision, recall and F‐measure. The authors present a case study where GitHub is used to show the effectiveness of authors’ approach. Authors’ model can recommend relevant human resources (HR) to project leaders who have participated in similar projects. The comparative experiment results show that the planned approach will effectively solve the sparseness drawback and produce suitable coverage rate and recommendation quality.

Summary Although the importance of models continuously grows in software development, common development approaches are less able to integrate the automatic management of model integrity into the development process. These critically important constraints may ensure the coherence of models in the evolution process to prevent manipulations that could violate defined constraints on a model. This paper proposes an integrity framework in the context of model‐driven architecture to achieve sufficient structural code coverage at a higher program representation level than machine code. Our framework offers to propagate the modifications from a platform‐independent specification to the corresponding test template model while keeping the consistency and integrity constraints after system evolution. To examine the efficiency of the proposed framework, a quantitative analysis plan is evaluated based on two experimental case studies. In addition, we propose coverage criteria for integrity regression testing (IRT), derived from logic coverage criteria that apply different conceptual levels of testing for the formulation of integrity requirements. The defined criteria for IRT reduce the inherent complexity and cost of verifying complex design changes in regression testing while keeping the fault detection capability with respect to the changes. The framework aims to keep pace with IRT in a formal way. The framework can solve a number of restricted outlooks in model integrity and some limiting factors of incremental maintenance and retesting. The framework satisfies several valuable quality attributes in software testing, such as safety percentage, precision, abstract fault detection performance measurable coverage level, and generality.

Cloud computing has changed the traditional large‐scale computational environment by making computing resources available on a pay‐per‐use basis and provides a new direction for network‐based applications by enabling sharing of services. The encouraging of cloud computing is to aggregate heterogeneous distributed sources to solve complicated industrial and scientific issues. The highest concentration of cloud‐computing systems is about sharing resources in a large‐scale multi‐organisational cooperation and their usage in new applications. To achieve this, an efficient scheduling system is a vital part for cloud computing. The dynamic and heterogeneous nature of cloud sources lead to the increased complexity of scheduling algorithms. Therefore, deterministic algorithms may not have enough efficiency to solve this issue. In this study, a new solution is presented to improve dynamic scheduling in cloud environments by combining greedy and max–min scheduling methods. The most important features of the proposed method include reduction of completion of the last task, reduction of total waiting time, observing of load balance and back up of data dynamic operations. The results of the authors’ simulations show performance improvement in comparison with greedy and max–min algorithms.

Abstract The increase in complexity and rate of technological changes in modern software development has led to a demand for systematic methods that raise the abstraction level for system maintenance and regression testing. Model‐driven development (MDD) has promised to reduce extra coding efforts in software maintenance activities using traceable change management. The research described in this paper presents a Z‐notation‐based framework, called delta‐based regression testing (DbRT), for formal modeling of regression testing in the context of MDD. The framework proposes to propagate the changes from a software specification to testing artifacts in order to preserve consistency after system evolution. Also, an effective delta‐based selection technique is provided for regression testing at the platform‐independent level. The framework is further enriched by introducing a new category of coverage patterns for DbRT. Complex coverage patterns can be defined using a declarative query language syntax for examining the adequacy of regression testing. Finally, an implementation technique and an analysis plan are provided to assess the effectiveness of the proposed framework. The assessment process is expected to be beneficial to both the platform‐independent and platform‐specific level of DbRT by identifying the desired coverage according to available testing resources. Copyright © 2015 John Wiley & Sons, Ltd.