Silvia Giannini

On the convergence of the max-consensus protocol with asynchronous updates

In this paper, we present new theoretical results on the convergence of max-consensus protocols for asynchronous networks. The analysis is carried out exploiting well-established concepts in the field of partially asynchronous iterative algorithms and of analytic synchronization. As a main result, we propose a theoretical setting to prove the convergence of the asynchronous max-consensus protocol. Moreover, we provide an upper bound on the convergence time of the max-consensus protocol in asynchronous networks.

Asynchronous Max-Consensus Protocol With Time Delays: Convergence Results and Applications

This paper deals with the analysis of the convergence properties of the max-consensus protocol in presence of asynchronous updates and bounded time delays on directed static networks. The work is motivated by real-world applications in distributed decision-making systems, for which max-consensus is an effective paradigm. The main result of this paper is that the strongly connectedness of the directed communication network is a sufficient condition for the asynchronous max-consensus protocol to let a distributed system converge in finite time. Implementation issues are also taken into account, by complementing the theoretical analysis with the definition of a mechanism to detect convergence in a distributed fashion. Finally, a numerical example is given, highlighting both the issues related to the failure of synchronous protocols applied to asynchronous settings and the effectiveness of the proposed asynchronous framework.

Asynchronous consensus-based distributed target tracking

This paper addresses the problem of distributed target tracking, performed by a network of agents which update their local estimates asynchronously. The proposed solution extends and improves an existing consensus-based distributed target tracking framework to cope with real-world settings in which each agent is driven by a different clock. In the consensus-based target tracking framework, it is assumed that only a few agents can actually measure the target state at a given time, whereas the remainder is able to perform a model-based prediction. Subsequently, an algorithm based on max-consensus makes all the agents agree, in finite time, on the best available estimate in the network. The limitations imposed by the assumption of synchronous updates of the network nodes are here overcome by the introduction of the concept of asynchronous iteration. Moreover, an event-based approach makes for the lack of a common time scale at the network level. Furthermore, the synchronous scenario can be derived as a special case of the asynchronous setting. Finally, numerical simulations confirm the validity of the approach.

Large scale skill matching through knowledge compilation

We present a logic-based framework for automated skill matching, able to return a ranked referral list and the related ranking explanation. Thanks to a Knowledge Compilation approach, a knowledge base in Description Logics is translated into a relational database, without loss of information. Skill matching inference services are then efficiently executed via SQL queries. Experimental results for scalability and turnaround times on large scale data sets are reported, confirming the validity of the approach.

Defining and computing Least Common Subsumers in RDF

Several application scenarios in the Web of Data share the need to identify the commonalities between a pair of RDF źresources. Motivated by such needs, we propose the definition and the computation of Least Common Subsumers (LCSs) in RDF. To this aim, we provide some original and fundamental reformulations, to deal with the peculiarities of RDF. First, we adapt a few definitions from Graph Theory to paths and connectedness in RDF-graphs. Second, we define rootedRDF-graphs ( r -graphs), in order to focus on a particular resource inside an RDF-graph. Third, we change the definitions of LCSs originally set up for Description Logics to r -graphs. According to the above reformulations, we investigate the computational properties of LCS in RDF, and find a polynomial-time characterization using a form of graph composition. This result remarkably distinguishes LCSs from Entailment in RDF, which is an NP-complete graph matching problem. We then devise algorithms for computing an LCS. A prototypical implementation works as a proof-of-concept for the whole approach in three application scenarios, and shows usefulness and feasibility of our proposal. Most of our examples are taken directly from real datasets, and are fully replicable thanks to the fact that the choice about which triples are selected for the computation is made explicit and flexible.

RDF Data Clustering

The Web is evolving from a Web of Documents to a Web of Data. Meanwhile, the development of Semantic Web applications opens the way for addressing complex information needs. In this scenario, clustering is identified as a crucial task for semantic mashups. After a thorough review of RDF clustering techniques, the paper addresses the open issues within the efficient exploitation of the knowledge contained in RDF data sources. Then, first promising attempts in exploring the applicability of community detection algorithms for RDF clustering are reported.

Coverage-aware distributed target tracking for mobile sensor networks

In this paper, we deal with the problem of distributed target tracking with heterogeneous mobile robots. We extend the static sensor network framework, presented by some of the authors in [1], introducing a cooperative motion control algorithm. In particular, each mobile node has limited sensing range and can estimate (either by measuring or by predicting) the target position. Then, a totally distributed algorithm exploits a suitable max-consensus protocol to reach a global agreement on the best estimate available in the network at a given time instant. The algorithm is based on the spread of a perception confidence value, locally computed by every single node and related to the Fisher Information, and requires a connected network to converge. Finally, a control law based on artificial potential fields allows the agents to move in order to improve the algorithm performance. In particular, the distributed motion control algorithm is capable to trade off between different requirements: to reduce the distance between the agent and the target to improve the accuracy of the measurement, to maintain the network connectivity, to perform collision avoidance, and to maintain a desidered coverage of the sensed area. Extensive simulation results are provided to confirm the suitability of the approach.

A Logic-Based Approach to Named-Entity Disambiguation in the Web of Data

Semantic annotation aims at linking parts of rough data (e.g., text, video, or image) to known entities in the Linked Open Data (LOD) space. When several entities could be linked to a given object, a Named-Entity Disambiguation (NED) problem must be solved. While disambiguation has been extensively studied in Natural Language Understanding (NLU), NED is less ambitious—it does not aim to the meaning of a whole phrase, just to correctly link objects to entities—and at the same time more peculiar since the target must be LOD-entities. Inspired by semantic similarity in NLU, this paper illustrates a way to solve disambiguation based on Common Subsumers of pairs of RDF resources related to entities recognized in the text. The inference process proposed for resolving ambiguities leverages on the DBpedia structured semantics. We apply it to a TV-program description enrichment use case, illustrating its potential in correcting errors produced by automatic text annotators (such as errors in assigning entity types and entity URIs), and in extracting a description of the main topics of a text in form of commonalities shared by its entities.

Embedding semantics in human resources management automation via SQL

Among enterprise business processes, those related to HR management are characterized by conflicting issues: on one hand, the peculiarities of intellectual capital ask for rather expressive representation languages to convey as many facets as possible; on the other hand, such processes deal with huge amounts of resources to be managed. For handling HR management tasks, our approach combines the representation power of a logical language with the information processing efficiency of a DBMS. It has been implemented in a fully functioning platform, I.M.P.A.K.T., that we present here highlighting its peculiarities for three relevant business processes: skill matching, task/team composition and company core competence identification.

A deductive approach to the identification and description of clusters in linked open data

We propose an approach for inferring clusters in collections of RDF resources based on the features shared by their descriptions. The approach grounds on an algorithm for computing Common Subsumers in RDF proposed in a previous research work. The clustering service introduced here returns not only a possible partition of resources in a collection, but also a description of the knowledge shared within each cluster, in terms of (generalized) RDF triples.