Tranos Zuva

Diversity and Serendipity in Recommender Systems

The present age of digital information has presented a heterogeneous online environment which makes it a formidable mission for a noble user to search and locate the required online resources timely. Recommender systems were implemented to rescue this information overload issue. However, majority of recommendation algorithms focused on the accuracy of the recommendations, leaving out other important aspects in the definition of good recommendation such as diversity and serendipity. This results in low coverage, long-tail items often are left out in the recommendations as well. In this paper, we present and explore a recommendation technique that ensures that diversity, accuracy and serendipity are all factored in the recommendations. The proposed algorithm performed comparatively well as compared to other algorithms in literature.

A context-aware proactive recommender system for tourist

With the fast growth in Information Communication Technology (ICT), the aspect of Recommender system have proved to be a great tool to /finding appropriate items/objects for users. Traditional user request response pattern are usually used to make these recommendations, adding context information and integrating proactivity to the process of recommending items/objects has improved recommender system so far. Context-Aware Proactive Recommender system can be of great assistance in giving appropriate recommendation in mobile devices. Therefore this study proposes a Context-aware proactive recommender system for tourists, it will also include contextual factors such as time, location and weather of the user and how a particular item satisfies users need, using Multi-criteria Collaborative filtering to make suggestions more efficient and accurate. Experiment will be conducted and the performance evaluation of the proposed system will be analysed to measure the efficiency and efficacy of the proposed techniques.

Towards the Adoption of Software Engineering Principles for Assessing and Ensuring the Reliability of Digital Forensic Tools

The swift mutative nature of digital evidence generally makes it open to challenge from usually the disadvantaged party during an incident proceeding. The fairness of the eventual ruling passed, after considering evidence, to the involved parties is important. This brings about the meticulousness with which legal systems admit digital evidence into the facts required to decide lawsuits. In response, digital forensics practitioners employ standardized methodologies that stress and guide the preservation of the integrity of digital evidence. Validation of software used at various phases of a digital forensic investigation is an important procedure within such methodologies. This paper explores the possibility of demonstrating the reliability of digital forensics software by the methodology with which they were developed. Various software engineering principles are analyzed for strengths and limitations, to examine how reliable the evidence produced by software that implement such principle may be regarded. The DESMET feature analysis and benchmarking evaluation methods are proposed to determine the most appropriate set of software development principles for digital forensics tools, as far as ensuring their non-damaging interactions with evidence is concerned. The paper explains the benefits, limitations and concepts behind the feasibility of this tool validation technique, as the actual evaluations and demonstration are not yet in place.

Comprehensive Diversity in Recommender Systems

The present age of digital information has presented a heterogeneous online environment which makes it a formidable mission for a noble user to search and locate the required online resources timely. Recommender systems were implemented to rescue this information overload issue. However, majority of recommender systems focus on the accuracy of the recommendations, leaving out other important aspects in the definition of good recommendation such as diversity and serendipity. This results in low coverage and long-tail items are often left out in the recommendations. In this paper, we present and explore a recommendation technique that ensures that comprehensive diversity is also factored-in in the recommendations. The algorithm adopts the second line of recommendation improvement whereby a recommendation list is re-ranked in such a way that it would include long-tail items. The results showed that the proposed algorithm is capable of giving a balanced list of recommendations in terms of accuracy and diversity.

Overview of Performance Evaluation of Keyword Search Utilities in Forensic/E-Discovery Software

Digital forensics has been modeled into a number of stages, which include examination and analysis. Keyword search is a popular tactic used by investigators during evidence examination and analysis. However, the belief that the success of forensic analysis depends on the examiner’s knowledge and experience has a strong hold in the digital forensic domain. It does imply the adequate awareness of the capabilities and limitations of the tools used by the examiner. Keyword search enables the examiner to quickly locate the existence of data items related to a case. This reduces investigation duration and eases the investigation process. This paper discusses the concepts of keyword search and the various keyword search techniques available. It highlights the algorithms on which they are based. In addition to the overview of, and argument for thorough understanding and evaluation of this technique in forensic utilities, this article also provides evaluation procedures to serve as direction for future evaluation/validation studies to ensure examiners know just how much to trust their software, as far as keyword searching is concerned.

Open issues in cybercriminal profiling

The ubiquity of cyber-incidents across individual and organizational realms suggests the inability of stakeholders to contain its pervasiveness. Reports across law enforcement, industry, and academia have corroborated this suggestion, with estimations of more explosive figures in the next couple of years. Insufficient investigative techniques, amongst others, have been identified as the stumbling block to effectively containing cybercrimes. Criminal profiling has recently been suggested by researchers as an investigative technique that could bring the immense value it has rendered in traditional crime investigations to cybercrime investigations. However, there have been a lot of difficulties encountered in the bid to apply this technique to cybercrime investigations, and researchers generally prefer to use more straightforward techniques. This paper seeks to highlight the technicalities of cybercrime that make the application of criminal profiling to its investigations such a difficult and dreadful undertaking. Recommendations are also provided about a probable solution to encourage more activity in this wanting area.

Hybrid Service Discovery Approach for Peer-to-peer Web Service Systems

Efficient service discovery is an essential task in distributed systems. Proliferation of web services has made this task difficult and challenging over the years. One of such challenges is the need to exhaustively search through all the services in the repositories to discover a required service. Another challenge is the huge number of irrelevant services returned during service discovery. The approach introduced in this work employs a clustering technique for the purpose of reducing the size of the search space and eliminating irrelevant services. In cases where the query is not satisfied within a super-node, an agent is activated to search and learn the traversed nodes to the required service. The performance of this approach is evaluated against two other approaches for service discovery. The results show a better performance in our approach over the other two approaches.

Survey of smart city initiatives towards urbanization

The smart city concept is a recent phenomenon that seems to be drawing so much attention from both academia and industry. With a rapid and alarming influx of migrants into urban areas the service delivery, traffic congestion to and fro the city, air pollution, crime rate, energy crisis (load shedding), standard of living has become a burning issue for both city managers and management. This event has triggered the need for smart cities initiates as a solution to remedy the situation. Although a number of pilot projects seem to be showing sustainability and great success in Europe, Asia and United states, similar projects (initiatives) need to be carried out or tested particularly in developing countries. This paper provides a survey into smart cities initiatives towards urbanization challenges. Open issues and challenges are discussed, and then the impact of smart cities on science, technology and society are highlighted.

Is there a room for security and privacy in IoT

Internet of things (IoT) continues to the draw attention of academics and researchers across the globe, since it represents the future of ubiquitous Computing. This is triggered by number of both digital and physical objects connecting to each other using ICT technologies, platforms and the internet to facilitate the whole processes of connectivity and services provision. With this massive connectivity of objects and devices forming the IoT, comes a great responsibility in terms of confronting new sets of challenges ranging from IoT security threats, ethics and privacy. In this paper we present the survey into the security aspect of IoT.

Traffic flow prediction in cloud computing

Cloud computing provides improved and simplified IT management and maintenance capabilities through central administration of resources, companies of all shapes and sizes are adapting to this new technology. In the absence of an effective prediction tools of cloud computing traffic then allocation of resources to clients will be ineffective thus driving away cloud computing users. In this paper we propose auto-regressive integrated moving average (ARIMA) and artificial neural networks (ANN) as prediction tools for cloud computing traffic. The results show that ARIMA performs better than ANN in predicting cloud computing traffic. For future work we propose to investigate the use of a hybrid of the two in predicting cloud computing traffic.