Claudia Marinica

Ontology-Based Data Integration System for Conservation-Restoration Data (OBDIS-CR)

In this paper, we present a semantic based approach for integrating heterogeneous data sources, and we implement our approach on the scientific domain of Cultural Heritage Conservation-Restoration. First, we evaluate the context and the current needs; the observations confirmed the critical need of a data integration system taking advantage of semantics. Second, we describe the process that we followed in order to build an ontology dedicated to provide a unified understanding of the data. Third, we describe the ontology-based mediator system designed for effectively querying the heterogeneous distributed sources.

Mining Frequent Itemsets with Vertical Data Layout in MapReduce

Frequent itemset mining is a Data Mining technique aiming to generate from a dataset new and interesting information under the form of sets of items. Several algorithms were already proposed, and successfully implemented and used such as Apriori, FP-Growth and Eclat, along with numerous improvements. These algorithms deal with two types of data layouts: horizontal and vertical; the former corresponds to the traditional layout (the individuals as rows and the items as columns) and it is more used due to its facility, but the latter brings important computation reductions. The standard frequent itemset mining algorithms have a high computational complexity and, given the available massive datasets, new approaches were proposed in the literature implementing mining algorithms in parallel, distributed, and lately Cloud Computing paradigms.

Mining Interesting Disjunctive Association Rules from Unfrequent Items

In most approaches to mining association rules, interestingness relies on frequent items, i.e., rules are built using items that frequently occur in the transactions. However, in many cases, data sets contain unfrequent items that can reveal useful knowledge that most standard algorithms fail to mine. For example, if items are products, it might be that each of the products \(p_1\) and \(p_2\) does not sell very well (i.e., none of them appears frequently in the transactions) but, that selling products \(p_1\) or \(p_2\) is frequent (i.e., transactions containing \(p_1\) or \(p_2\) are frequent). Then, assuming that \(p_1\) and \(p_2\) are similar enough with respect to a given similarity measure, the set \(\{p_1, p_2\}\) can be considered for mining relevant rules of the form \(\{p_1, p_2\} \rightarrow \{p_3, p_4\}\) (assuming that \(p_3\) and \(p_4\) are unfrequent similar products such that \(\{p_3,p_4\}\) is frequent), meaning that most of customers buying \(p_1\) or \(p_2\), also buy \(p_3\) or \(p_4\). The goal of our work is to mine association rules of the form \(D_1 \rightarrow D_2\) such that \((i)\) \(D_1\) and \(D_2\) are disjoint homogeneous frequent itemsets made up with unfrequent items, and \((ii)\) the support and the confidence of the rule are respectively greater than or equal to given thresholds. The main contributions of this paper towards this goal are to set the formal definitions, properties and algorithms for mining such rules.

Personal Networks of Scientific Collaborators: A Large Scale Experimental Analysis of Their Evolution

When an individual joins an Online Social Network (OSN), he creates connections by interacting with the other users directly or indirectly and forms its own Online Personal Network (OPN). These OPNs are not static, but they evolve over time as new people join or quit them and as new relationships are established or old ones broken. Understanding how OPNs are evolving is still missing in the current literature, while OSNs’ evolution was widely addressed and many models were proposed. In this paper, we propose to fill this gap by performing an experimental analysis over a large set of real OPNs by the mean of the computation of metrics that characterize their structure. We examine how these metrics behave when the OPNs change over time in order to discover the properties driving the evolution of their structure, which can help in providing evolution models dedicated to OPNs.

Mining Frequent and Homogeneous Closed Itemsets

It is well known that when mining frequent itemsets from a transaction database, the output is usually too large to be effectively exploited by users. To cope with this difficulty, several forms of condensed representations of the set of frequent itemsets have been proposed, among which the notion of closure is one of the most popular.