Maria Malek - Complex network analysis

Overlapping community detection using core label propagation algorithm

A methodology for multilayer networks analysis in the context of open and private data: biological application

Social-semantic recommender systems

We consider the basic problem in recommender systems that consists in identifying a set of users to whom a given item is to be recommended. In practice recommender systems are run against huge sets of users, and the problem is then to avoid scanning the whole user set in order to produce the recommendation list. In order to cope with problem, we consider that users are connected through a complex network and that a taxonomy over the items has been defined. These two kinds of information, respectively called social and semantic information, are combined in our approach in order to compute recommendation lists by visiting a limited part of the complex network. In doing so, our contribution is to propose recommendation algorithms that outperform existing approaches, while producing recommendation lists of similar accuracy.

In our experiments, we use two real data sets to test our algorithms, namely Amazon.com and Movie- Lens data sets, and we compare our algorithms with the standard item-based collaborative filtering and hybrid recommendation algorithms. Our results show a satisfying accuracy values and a very significant improvement of performance, by exploring a small part of the graph instead of exploring all the graph as the classical searching methods do.

(paper)

Overlapping community detection using core label propagation algorithm

The community detection in complex networks has become a major field of research. Disjoint community detection deals often with getting a partition of nodes where every node belongs to only one community. However, in social networks, individuals may belong to more than one community such as in co-purchasing field, a co-authorship of scientist papers or anthropological networks. We propose a method to find overlapping communities from pre-computed disjoint communities obtained by using the core detection label propagation. The algorithm selects candidates nodes for overlapping and uses belonging functions to decide the assignment or not of a candidate node to each of its neighbours communities. we propose and experiment in this paper several belonging functions, all based on the topology of the communities. These belonging functions are either based on global measures which are the density and the clustering coefficient or on average node measures which are the betweenness and the closeness centralities. We expose then a new similarity measure between two covers regarding the overlapping nodes. The goal is to assess the similarity between two covers that overlap several communities. We finally propose a comparative analysis with the literature algorithms.

(paper)

A methodology for multilayer networks analysis in the context of open and private data: biological application

Recently, an increasing body of work investigates networks with multiple types of links. Variants of such systems have been examined decades ago in disciplines such as sociology and engineering, but only recently have they been unified within the framework of multilayer networks. In parallel, many aspects of real systems are increasingly and routinely sensed, measured and described, resulting in many private, but also open data sets. In many domains publicly available repositories of open data sets constitute a great opportunity for domain experts to contextualise their privately generated data compared to publicly available data in their domain. We propose a methodology for multilayer network analysis in order to provide domain experts with measures and methods to understand, evaluate and complete their private data by comparing and/or combining them with open data when both are modelled as multilayer networks. We illustrate our methodology through a biological application where interactions between molecules are extracted from open databases and modelled by a multilayer network and where private data are collected experimentally. This methodology helps biologists to compare their private networks with the open data, to assess the connectivity between the molecules across layers and to compute the distribution of the identified molecules in the open network. In addition, the shortest paths which are biologically meaningful are also analysed and classified.

(paper)