@ntnu.edu
Research Associate, Department of Computer Science
Norwegian University of Science and Technology
Scopus Publications
Scholar Citations
Scholar h-index
Scholar i10-index
Georgios K. Pitsilis, Heri Ramampiaro, and Helge Langseth
Springer Science and Business Media LLC
This paper addresses the important problem of discerning hateful content in social media. We propose a detection scheme that is an ensemble of Recurrent Neural Network (RNN) classifiers, and it incorporates various features associated with userrelated information, such as the users’ tendency towards racism or sexism. These data are fed as input to the above classifiers along with the word frequency vectors derived from the textual content. Our approach has been evaluated on a publicly available corpus of 16k tweets, and the results demonstrate its effectiveness in comparison to existing state of the art solutions. More specifically, our scheme can successfully distinguish racism and sexism messages from normal text, and achieve higher classification quality than current state-of-the-art algorithms.
Wei Wang, Jiqiang Liu, Georgios Pitsilis, and Xiangliang Zhang
Elsevier BV
Abstract Anomaly intrusion detection in big data environments calls for lightweight models that are able to achieve real-time performance during detection. Abstracting audit data provides a solution to improve the efficiency of data processing in intrusion detection. Data abstraction refers to abstract or extract the most relevant information from the massive dataset. In this work, we propose three strategies of data abstraction, namely, exemplar extraction, attribute selection and attribute abstraction. We first propose an effective method called exemplar extraction to extract representative subsets from the original massive data prior to building the detection models. Two clustering algorithms, Affinity Propagation (AP) and traditional k-means, are employed to find the exemplars from the audit data. k-Nearest Neighbor (k-NN), Principal Component Analysis (PCA) and one-class Support Vector Machine (SVM) are used for the detection. We then employ another two strategies, attribute selection and attribute extraction, to abstract audit data for anomaly intrusion detection. Two http streams collected from a real computing environment as well as the KDD’99 benchmark data set are used to validate these three strategies of data abstraction. The comprehensive experimental results show that while all the three strategies improve the detection efficiency, the AP-based exemplar extraction achieves the best performance of data abstraction.
Georgios Pitsilis and Wei Wang
Elsevier BV
Used the collaborative tagging idea to produce personalized recommendations.Attempted utilization of the power of taxonomies through clustering of annotations.Experimental evaluation of the model using data from public annotation system, citeUlike.The approach improved prediction quality without compromising computational efficiency. Social bookmarking and tagging has emerged a new era in user collaboration. Collaborative Tagging allows users to annotate content of their liking, which via the appropriate algorithms can render useful for the provision of product recommendations. It is the case today for tag-based algorithms to work complementary to rating-based recommendation mechanisms to predict the user liking to various products. In this paper we propose an alternative algorithm for computing personalized recommendations of products, that uses exclusively the tags provided by the users. Our approach is based on the idea of using the semantic similarity of the user-provided tags for clustering them into groups of similar meaning. Afterwards, some measurable characteristics of users' Annotation Competency are combined with other metrics, such as user similarity, for computing predictions. The evaluation on data used from a real-world collaborative tagging system, citeUlike, confirmed that our approach outperforms the baseline Vector Space model, as well as other state of the art algorithms, predicting the user liking more accurately.
Xiang-Liang Zhang, Tak Man Desmond Lee, and Georgios Pitsilis
Springer Science and Business Media LLC
Recommender systems (RS) have been found supportive and practical in e-commerce and been established as useful aiding services. Despite their great adoption in the user communities, RS are still vulnerable to unscrupulous producers who try to promote their products by shilling the systems. With the advent of social networks new sources of information have been made available which can potentially render RS more resistant to attacks. In this paper we explore the information provided in the form of social links with clustering for diminishing the impact of attacks. We propose two algorithms, CluTr and WCluTr, to combine clustering with \\trust" among users. We demonstrate that CluTr and WCluTr enhance the robustness of RS by experimentally evaluating them on data from a public consumer recommender system Epinions.com.
Mohamed El-Hadedy, Georgios Pitsilis, and Svein J. Knapskog
IEEE
Many electronic content providers today like Flickr and Google, offer space to users to publish their electronic media(e.g. photos and videos) in their cloud infrastructures so that they can be publicly accessed. Features like including other information, such as keywords or owner information into the digital material is already offered by existing providers. Despite the useful features made available to users by such infrastructures, the authorship of the published content is not protected against various attacks such as compression. In this paper we propose a robust scheme that uses digital invisible watermarking and hashing to protect the authorship of the digital content and provide resistance against malicious manipulation of multimedia content. The scheme is enhanced by an algorithm called MMBEC, that is an extension of an established scheme MBEC towards higher resistance.
Pern Hui Chia and Georgios Pitsilis
Information Processing Society of Japan
The majority of recommender systems predict user preferences by relating users with similar attributes or taste. Prior research has shown that trust networks improve the accuracy of recommender systems, predominantly using algorithms devised by individual researchers. In this work, omitting any specific trust inference algorithm, we investigate how useful it might be if explicit trust relationships are used to select the best neighbors or predictors, to generate accurate recommendations. We conducted a series of evaluations using data from Epinions.com, a popular collaborative reviewing system. We find that, for highly active users, using trusted sources as predictors does not give more accurate recommendations compared to the classic similarity-based collaborative filtering scheme, except in improving the precision to recommend items that are of users' liking. This cautions against the intuition that inputs from trusted sources would always be more accurate or helpful. The use of explicit trust links, however, provides a slight gain in prediction accuracy when it comes to the less active users. These findings highlight the need and potential to adapt the use of trust information for different groups of users, besides to better understand trust when employing it in the recommender systems. Parallel to the trust criterion, we also investigated the effects of requiring the candidate predictors to have an equal or higher experience level.
Georgios Pitsilis, Xiangliang Zhang, and Wei Wang
Springer Berlin Heidelberg
In this work, we explore the benefits of combining clustering and social trust information for Recommender Systems. We demonstrate the performance advantages of traditional clustering algorithms like k-Means and we explore the use of new ones like Affinity Propagation (AP). Contrary to what has been used before, we investigate possible ways that social-oriented information like explicit trust could be exploited with AP for forming clusters of high quality. We conducted a series of evaluation tests using data from a real Recommender system Epinions.com from which we derived conclusions about the usefulness of trust information in forming clusters of Recommenders. Moreover, from our results we conclude that the potential advantages in using clustering can be enlarged by making use of the information that Social Networks can provide.
Wei Wang, Xiangliang Zhang, and Georgios Pitsilis
Springer Berlin Heidelberg
High speed of processing massive audit data is crucial for an anomaly Intrusion Detection System (IDS) to achieve real-time performance during the detection. Abstracting audit data is a potential solution to improve the efficiency of data processing. In this work, we propose two strategies of data abstraction in order to build a lightweight detection model. The first strategy is exemplar extraction and the second is attribute abstraction. Two clustering algorithms, Affinity Propagation (AP) as well as traditional k-means, are employed to extract the exemplars, and Principal Component Analysis (PCA) is employed to abstract important attributes (a.k.a. features) from the audit data. Real HTTP traffic data collected in our institute as well as KDD 1999 data are used to validate the two strategies of data abstraction. The extensive test results show that the process of exemplar extraction significantly improves the detection efficiency and has a better detection performance than PCA in data abstraction.
Georgios Pitsilis and Pern Hui Chia
Springer Berlin Heidelberg
Recommender systems have evolved during the last few years into useful online tools for assisting the daily e-commerce activities. The majority of recommender systems predict user preferences relating users with similar taste. Prior research has shown that trust networks improve the performance of recommender systems, predominantly using algorithms devised by individual researchers. In this work, omitting any specific trust inference algorithm, we investigate how useful it might be if explicit trust relationships (expressed by users for others) are used to select the best neighbours (or predictors), for the provision of accurate recommendations. We conducted our experiments using data from Epinions.com, a popular recommender system. Our analysis indicates that trust information can be helpful to provide a slight performance gain in a few cases especially when it comes to the less active users.
Georgios Pitsilis
Springer Berlin Heidelberg
Trust has been explored by many researchers in the past as a solution for assisting the process of recommendation production. In this work we are examining the feasibility of building networks of trusted users using the existing evidence that would be provided by a standard recommender system. As there is lack of models today that could help in finding the relationship between trust and similarity we build our own that uses a set of empirical equations to map similarity metrics into Subjective Logic trust. In this paper we perform evaluation of the proposed model as being a part of a complete recommender system. Finally, we present the interesting results from this evaluation that shows the performance and benefits of our trust modeling technique as well as its impact on the user community as it evolves over time.
Georgios Pitsilis and Lindsay F. Marshall
Springer US
In this paper we present novel techniques for modeling trust relationships that can be used in recommender systems. Such environments exist with the voluntary collaboration of the community members who have as a common purpose the provision of accurate recommendations to each other. The performance of such systems can be enhanced if the potential trust between the members is properly exploited. This requires that trust relationships are appropriately established between them. Our model provides a link between the existing knowledge, expressed in similarity metrics, and beliefs which are required for establishing a trust community. Although we explore this challenge using an empirical approach, we attempt a comparison between the alternative candidate formulas with the aim of finding the optimal one. A statistical analysis of the evaluation results shows which one is the best. We also compare our new model with existing techniques that can be used for the same purpose.
Georgios Pitsilis and Panayiotis Periorellis
IEEE
Peer-to-peer networks exist with the volunteering cooperation of various entities on the Internet. Their self-structure nature has the important characteristic that they make no use of central entities to run as coordinators and the benefits of this cooperation can be enjoyed equally by all the members of the community with the assumption that they all make right use of the protocol. In this study we examine what the consequences on the community are in the case of existence of misbehaving nodes which can abuse the network resources for their personal benefit and we also analyze the cost and the benefit of some proposed solution that could be used in order to bring into account the above problem
Georgios Pitsilis and Lindsay Marshall
IEEE
In this paper we present a trust-oriented method that can be used when building P2P recommender systems. We discuss its benefits in comparison to centralized solutions, its requirements, its pitfalls and how these can be overcome. We base the formation of trust on evidential reasoning and designed with ease of adoption by existing infrastructures in mind. The paper includes a first analysis of performance based on a simulation used to investigate the impact on scalability and thus show the applicability of the protocol
Georgios Pitsilis, Panayiotis Periorellis, and Lindsay Marshall
Springer Berlin Heidelberg
Unstructured P2P networks, despite having good characteristics such as the nonexistence of a single point of failure, the high levels of anonymity in the search operations and the exemplary dependability, have been found to be much less scalable than first expected. The flooding protocol, which is used for the discovery of peers and for the main operation of searching, seems to be responsible for this weakness. The adoption of some major improvements, such as the distinction between Leaf-nodes and Ultra-Peers, has partially overcome the scalability problems, but there is still a need for further optimization. Our proposed idea, aims to improve the effectiveness of the hierarchical scheme by applying some new criteria in the selection of potentially promotable nodes.
Georgios Pitsilis and Lindsay Marshall
Springer Berlin Heidelberg
In this paper we propose a method that can be used to avoid the problem of sparsity in recommendation systems and thus to provide improved quality recommendations. The concept is based on the idea of using trust relationships to support the prediction of user preferences. We present the method as used in a centralized environment; we discuss its efficiency and compare its performance with other existing approaches. Finally we give a brief outline of the potential application of this approach to a decentralized environment.
Georgios Pitsilis
Springer Berlin Heidelberg
Peer-to-Peer information sharing environments have gained recognition and popularity during the recent years. In spite of the useful characteristics they provide in the ways that the participants can collaborate, the issue of quality preservation in the shared material has not been addressed yet. The lack of appropriate mechanisms and policies to evaluate the participants has sown fears that the overall popularity of the services will be affected. The nature of atomistic p2p models, where survivability is based on the idea of self-organization into communities could be the basis of a solution to the quality problem build-up by the peers themselves. We consider that the deployment of an assessment scheme as a consultancy service based on a localized view of reputation could help the associated members of the peer-to-peer community in making their choices and thus in the provision of better services.