Giovanni Cozzolino

Detecting anomalies in Twitter stream for public security issues

Social networking services gain more often interest for research goals in several fields and applications thanks to the big amount of data that users daily post on them. Knowledge that has accumulated in the social sites enables to catch the reflection of real world events. In this work we present a general framework for event detection from Twitter. The framework aims to collect tweets related to a particular social event, in order to filter and classify those which can be relevant to detect malicious actions in Twitter communities. Relevant tweets are processed to raise an alert in case of anomaly within the collected set.

Overview of Digital Forensic Tools for DataBase Analysis

The number of digital devices that people use in everyday life has significantly increased. Since they have become an integral part of everyday life, they contain information that are often extremely sensitive. Modern devices use complex data structures to store data (heterogeneous media files, documents, GPS positions and SQLite databases, etc.), therefore, during a forensic investigation, it has been necessary the adoption of specialized acquisition and analysis tools.

Semantic Analysis of Social Data Streams

Social Networks Analysis has become a common trend among scholars and researchers worldwide. A great number of companies, institutions and organisations are interested in social networks data mining. Information published on many social networks, like Facebook, Twitter or Instagram constitute an important asset in many application fields, overall sentiment analysis, but also economics analysis, politics analysis and so on. Social networks analysis comprehends many disciplines and involves the application of different methodologies and techniques to define the criteria for generating the analytics, according to the purpose of the study. In this work, we focused on the semantic analysis of the content of textual information obtained from social media, aiming at extracting hot topics from social networks. We considered, as case study, reviews from the Yelp social network. The same methodology can be also applied for social and political opinion mining campaigns.

An Advanced Methodology to Analyse Data Stored on Mobile Devices

Nowadays computer and mobile devices, such as mobile phones, smartphones, smartwatches, tablets, etc., represent the multimedia diary of each of us. Thanks to technological evolution and the advent of an infinite number of applications, mainly aimed at socialization and entertainment, they have become the containers of an infinite number of personal and professional information. For this reason, optimizing the performance of systems able to detect intrusions (IDS - Intrusion Detection System) is a goal of common interest. This paper presents a methodology to classify hacking attacks taking advantage of the generalization property of neural networks. In particular, in this work we adopt the multilayer perceptron (MLP) model with the back-propagation algorithm and the sigmoidal activation function. We analyse the results obtained using different configurations for the neural network, varying the number of hidden layers and the number of training epochs in order to obtain a low number of false positives. The obtained results will be presented in terms of type of attacks and training epochs and we will show that the best classification is carried out for DOS and Probe attacks.

A MAS Model for Reaching Goals in Critical Systems

The exploitation of Cloud infrastructure in Big Data management is appealing because of costs reductions and potentiality of storage, network and computing resources. The Cloud can consistently reduce the cost of analysis of data from different sources, opening analytics to big storages in a multi-cloud environment. Anyway, creating and executing this kind of service is very complex since different resources have to be provisioned and coordinated depending on users’ needs. Orchestration is a solution to this problem, but it requires proper languages and methodologies for automatic composition and execution. In this work we propose a methodology for composition of services used for analyses of different Big Data sources: in particular an Orchestration language is reported able to describe composite services and resources in a multi-cloud environment.

Using Multilayer Perceptron in Computer Security to Improve Intrusion Detection

Nowadays computer and network security has become a major cause of concern for experts community, due to the growing number of devices connected to the network. For this reason, optimizing the performance of systems able to detect intrusions (IDS - Intrusion Detection System) is a goal of common interest. This paper presents a methodology to classify hacking attacks taking advantage of the generalization property of neural networks. In particular, in this work we adopt the multilayer perceptron (MLP) model with the back-propagation algorithm and the sigmoidal activation function. We analyse the results obtained using different configurations for the neural network, varying the number of hidden layers and the number of training epochs in order to obtain a low number of false positives. The obtained results will be presented in terms of type of attacks and training epochs and we will show that the best classification is carried out for DOS and Probe attacks.

Data Mining in Social Network

In this paper, we propose a novel data model for Multimedia Social Networks, i.e. particular social media networks that combine information on users belonging to one or more social communities together with the content that is generated and used within the related environments. The proposed model relies on the hypergraph data structure to capture and represent in a simple way all the different kinds of relationships that are typical of social media networks, and in particular among users and multimedia content. We also introduce some user and multimedia ranking functions to enable different applications. Finally, some experiments concerning effectiveness of the approach for supporting relevant information retrieval activities are reported and discussed.

Improving Results of Forensics Analysis by Semantic-Based Suggestion System

Nowadays, more than ever, digital forensics activities are involved in any criminal, civil or military investigation and they are primary to support cyber-security. Detectives use a many techniques and proprietary forensic software to analyze (copies of) digital devices, in order to discover hidden, deleted, encrypted, and damaged files or folders. Any evidence found is carefully analysed and documented in “finding reports” that are used during lawsuits. Forensics aim at discovering and analysing patterns of fraudulent activities. In this work, we propose a methodology that supports detectives in correlating evidences found by different forensic tools and we apply it to a framework able to semantically annotate data generated by forensics tools. Annotations enable more effective access to relevant information and enhanced retrieval and reasoning.

Detect and correlate information system events through verbose logging messages analysis

Detecting and tracking events from logging data is a critical element for security and system administrators and thus attracts more and more research efforts. However, there exists a major limitation in current processes of Event Logging analysis, related to the verbosity and language-dependence of messages produced by many logging systems. In this paper, a novel methodology was proposed to tackle this limitation by analysing event messages through a Natural Language Processing task in order to annotate them with semantic metadata. These metadata are further used to enable semantic searches or domain ontology population that help administrator to filter only relevant event and to correlate them for a prompt and efficient response and incident analysis.

Sentiment Analysis on yelp social network.

Social networks analysis is an emerging trend among scholars and researchers in the last years. A great number of companies are interested in social networks data mining. Data gathered from Facebook, Twitter or other social networks result to be very attractive in many application fields, like economics analysis, sentiment analysis, and politics analysis and so on. In our work, we focused on the analysis of the content of textual information obtained from the social media. Our investigation is finalized to extract hot topics in social network. We considered, as case study, reviews obtained from the social network Yelp.