Konstantinos Tserpes

Recommender Systems for Large-Scale Social Networks: A review of challenges and solutions

Abstract Social networks have become very important for networking, communications, and content sharing. Social networking applications generate a huge amount of data on a daily basis and social networks constitute a growing field of research, because of the heterogeneity of data and structures formed in them, and their size and dynamics. When this wealth of data is leveraged by recommender systems, the resulting coupling can help address interesting problems related to social engagement, member recruitment, and friend recommendations. In this work we review the various facets of large-scale social recommender systems, summarizing the challenges and interesting problems and discussing some of the solutions.

Predicting Edge Signs in Social Networks Using Frequent Subgraph Discovery

In signed social networks, users are connected via directional signed links that indicate their opinions about each other. Predicting the signs of such links is crucial for many real-world applications, such as recommendation systems. The authors mine patterns that emerge frequently in the social graph, and show that such patterns possess enough discriminative power to accurately predict the relationships among social network users. They evaluate their approach through an experimental study that comprises three large-scale, real-world datasets and show that it outperforms state-of-the art methods.

Computational workload prediction for grid oriented industrial applications: the case of 3D-image rendering

Grids are typically used for solving large-scale resource and computing intensive problems in science, engineering, and commerce as they seem to be cost-effective for industrial users. In order to be able to meet this requirement the software modules developed should be designed to meet the requisites for commercial business processes on the grid. In this paper we present a module for predicting computational workload of jobs assigned for execution on commercially exploited grid infrastructures. The module aims to identify the complexity of a given job and predict the workload that it is going to stress on the grid infrastructure. The prediction is achieved with the use of a trained artificial neural network, which has been implemented, with the use of the open source software package Joone. The approach has been implemented and validated within the framework of GRIA IST project for a specific industrial based application namely, 3D image rendering. The evaluation of the approach showed very promising results not only for the adoption of an open source package in a commercial application but also concerning the accuracy of the prediction and the benefit that it can provide in grids for business.

Effective QoS Monitoring in Large Scale Social Networks

Social Networking activities are still occupying the majority of the time that Internet users are spending in the Web. The generated content and social dynamics represent precious resources that everybody wishes to control. This scenario poses several challenges including the fact that different implementations, technologies, and formats are used to manage web content and social dynamics in heterogeneous, often antagonistic, Social Networking Sites. In order to master this heterogeneity the SocIoS project has defined an API that enables the aggregation of data and functionality made available by different Social Networking Sites APIs and their combination into complex and novel application workflows. However, the dependency on Social Networking Sites does not allow users of the SocIoS API to control the Quality of Service provided by the underlying platforms. In this paper we show how the QoSMONaaS (QoSMONitoring as a Service) component can be used to monitor and evaluate relevant metrics, such as availability and response time of the API calls, that are specified in the Service Level Agreement document. QoSMONaaS has been developed within the context of the SRT-15 project to implement a dependable (i.e. unbiased, reliable, and timely) monitoring of Quality of Service.

Semantically-enabled Intelligent Patient Recruitment in Clinical Trials

This paper presents and analyzes the data requirements within the clinical trial design process with particular focus on the selection of patients who are eligible to participate in the specified clinical study. The latter comprises an extremely time-consuming process which requires considerable budget and effort, whereas the resulting recruited subjects determine both the success of the clinical study and the validity of the clinical study results significantly. Hence, a novel approach based on Service Oriented Architecture mechanisms and the incorporation of a great number of ontologies enabling the semantic linking between clinical research and clinical care and the interpretation of numerous distributed heterogeneous data sources is presented and analyzed followed by the expected impact of its application in the clinical study design and implementation processes.

Employing traditional machine learning algorithms for big data streams analysis

A real-time trajectory prediction application is proposed.Performance vs. accuracy between big-data tailored solutions and traditional methods.Ignoring time element in time-series data can help in generalization of model. In this paper, we model the trajectory of sea vessels and provide a service that predicts in near-real time the position of any given vessel in 4, 10, 20 and 40 time intervals. We explore the necessary tradeoffs between accuracy, performance and resource utilization is explored given the large volume and update rates of input data. We start with building models based on well-established machine learning algorithms using static datasets and multi-scan training approaches and identify the best candidate to be used in implementing a single-pass predictive approach, under real-time constraints. The results are measured in terms of accuracy and performance and are compared against the baseline kinematic equations. Results show that it is possible to efficiently model the trajectory of multiple vessels using a single model, which is trained and evaluated using an adequately large, static dataset, thus achieving a significant gain in terms of resource usage while not compromising accuracy.

SocIoS API: A Data Aggregator for Accessing User Generated Content from Online Social Networks

Following the boost in popularity of online social networks, both enterprises and researchers looked for ways to access the social dynamics information and user generated content residing in these spaces. This endeavor, however, presented several challenges caused by the heterogeneity of data and the lack of a common way to access them. The SocIoS framework tries to address these challenges by providing tools that operate on top of multiple popular social networks allowing uniform access to their data. It provides a single access point for aggregating data and functionality from the networks, as well as a set of analytical tools for exploiting them. In this paper we present the SocIoS API, an abstraction layer on top of the social networks exposing operations that encapsulate the functionality of their APIs. Currently, the component provides support for seven social networks and is flexible enough to allow for the seamless addition of more.

A Classification of NoSQL Data Stores Based on Key Design Characteristics

Abstract Traditional Relational Database Management Systems are continuously being replaced by NoSQL data stores as a result of the growing demand for big data applications. The emergence of a large number of implementations of such like systems is a contributing indicator. This paper deals with the analysis of some key design characteristics of NoSQL systems and uses these for their characterization based on their capabilities. Furthermore, it highlights the relationship between NoSQL systems and cloud infrastructures and explains the impact that the existence of one has to the other.

Clustering Documents using the 3-Gram Graph Representation Model

In this paper we illustrate an innovative clustering method of documents using the 3-Gram graphs representation model and deducing the problem of document clustering to graph partitioning. For the latter we employ the kernel k-means algorithm. We evaluated the proposed method using the Test Collections of Reuters-21578, and compared the results using the Latent Dirichlet Allocation (LDA) Algorithm. The results are encouraging demonstrating that the 3-Gram graph method has much better Recall and F1 score but worse Precision. Further changes that will further improve the results are identified.

Evaluating Quality Provisioning Levels in Service Oriented Business Environments

This paper advocates the need for a mechanism that will allow the evaluation of the provided quality of service (QoS) by a service provider to a service customer in B2B service provisioning. Furthermore, this study goes on with presenting the feasibility of this mechanism by designing and testing a reference implementation that can be used in a service oriented architecture (SOA) environment which is able to support business application services. Experience gained in the frame of the NextGRID IST project that focused on the business perspectives of Grid computing and adopted SOA as its baseline architecture, has shown that such a mechanism is essential for enabling the economic viability of such large scale computing platforms.