Search

Abstract With real-world network systems typically comprising a large number of interactive components and inherently dynamic, Graph Continual Learning (GCL) has gained increasing popularity in recent years. Furthermore, most applications involve multiple entities and relationships with associated attributes, which has led to widely adopting Heterogeneous Information Networks (HINs) for capturing such rich structural and semantic meaning. In this context, we deal with the problem of learning multi-type node representations in a time evolving graph setting, harnessing the expressive power of Graph Neural Networks (GNNs). To this purpose, we propose a novel framework, named DyHANE—Dynamic Heterogeneous Attributed Network Embedding, which dynamically identifies a representative sample of multi-typed nodes as training set and updates the parameters of a GNN module, enabling the generation of up-to-date representations for all nodes in the network. We show the advantage of employing HINs on a data-incremental classification task. We compare the results obtained by DyHANE on a multi-step, incremental heterogeneous GAT model trained on a sample of changed and unchanged nodes, with the results obtained by either the same model trained from scratch or the same model trained solely on changed nodes. We demonstrate the effectiveness of the proposed approach in facing two major related challenges: (i) to avoid model re-train from scratch if only a subset of the network has been changed and (ii) to mitigate the risk of losing established patterns if the new nodes exhibit unseen properties. To the best of our knowledge, this is the first work that deals with the task of (deep) graph continual learning on HINs.

Abstract The migration of Twitter users to Mastodon following Elon Musk’s acquisition presents a unique opportunity to study collective behavior and gain insights into the drivers of coordinated behavior in online media. We analyzed the social network and the public conversations of about 75,000 migrated users and observed that the temporal trace of their migrations is compatible with a phenomenon of social influence, as described by a compartmental epidemic model of information diffusion. Drawing from prior research on behavioral change, we delved into the factors that account for variations of the effectiveness of the influence process across different Twitter communities. Communities in which the influence process unfolded more rapidly exhibit lower density of social connections, higher levels of signaled commitment to migrating, and more emphasis on shared identity and exchange of factual knowledge in the community discussion. These factors account collectively for 57% of the variance in the observed data. Our results highlight the joint importance of network structure, commitment, and psycho-linguistic aspects of social interactions in characterizing grassroots collective action, and contribute to deepen our understanding of the mechanisms that drive processes of behavior change of online groups.

Non-Fungible Tokens (NFTs) have emerged as the most representative application of blockchain technology in recent years, fostering the development of the Web3. Nonetheless, while the interest in NFTs rapidly boomed, creating unprecedented fervour in traders and creators, the demand for highly representative and up-to-date data to shed light on such an intriguing yet complex domain mostly remained unmet. To pursue this objective, we introduce a large collection of NFT transactions and associated metadata that correspond to trading operations between 2021 and 2023. Our developed dataset is the most extensive and representative in the NFT landscape to date, as it contains more than 70 M transactions performed by more than 6 M users across 36.3 M NFTs and 281 K collections. Moreover, this dataset boasts a wealth of metadata, including encoded textual descriptions and multimedia content, thus being suitable for a plethora of tasks relevant to database systems, AI, data science, Web and network science fields. This dataset represents a unique resource for researchers and industry practitioners to delve into the inner workings of NFTs through a multitude of perspectives, paving the way for unprecedented opportunities across multiple research fields.

Abstract Graph representation learning has become a topic of great interest and many works focus on the generation of high-level, task-independent node embeddings for complex networks. However, the existing methods consider only few aspects of networks at a time. In this paper, we propose a novel framework, named Co-MLHAN, to learn node embeddings for networks that are simultaneously multilayer, heterogeneous and attributed. We leverage contrastive learning as a self-supervised and task-independent machine learning paradigm and define a cross-view mechanism between two views of the original graph which collaboratively supervise each other. We evaluate our framework on the entity classification task. Experimental results demonstrate the effectiveness of Co-MLHAN and its variant Co-MLHAN-SA, showing their capability of exploiting across-layer information in addition to other types of knowledge.

Abstract Graph Neural Networks (GNNs) are powerful tools that are nowadays reaching state of the art performances in a plethora of different tasks such as node classification, link prediction and graph classification. A challenging aspect in this context is to redefine basic deep learning operations, such as convolution, on graph-like structures, where nodes generally have unordered neighborhoods of varying size. State-of-the-art GNN approaches such as Graph Convolutional Networks (GCNs) and Graph Attention Networks (GATs) work on monoplex networks only, i.e., on networks modeling a single type of relation among an homogeneous set of nodes. The aim of this work is to generalize such approaches by proposing a GNN framework for representation learning and semi-supervised classification in multilayer networks with attributed entities, and arbitrary number of layers and intra-layer and inter-layer connections between nodes. We instantiate our framework with two new formulations of GAT and GCN models, namely ML-GCN and ML-GAT, specifically devised for general, attributed multilayer networks. The proposed approaches are evaluated on an entity classification task on nine widely used real-world network datasets coming from different domains and with different structural characteristics. Results show that both our proposed ML-GAT and ML-GCN methods provide effective and efficient solutions to the problem of entity classification in multilayer attributed networks, being faster to learn and offering better accuracy than the competitors. Furthermore, results show how our methods are able to take advantage of the presence of real attributes for the entities, in addition to arbitrary inter-layer connections between the nodes in the various layers.

Abstract Open-source, Decentralized Online Social Networks (DOSNs) are emerging as alternatives to the popular yet centralized and profit-driven platforms like Facebook or Twitter. In DOSNs, users can set up their own server, or instance, while they can actually interact with users of other instances. Moreover, by adopting the same communication protocol, DOSNs become part of a massive social network, namely the Fediverse. Mastodon is the most relevant platform in the Fediverse to date, and also the one that has attracted attention from the research community. Existing studies are however limited to an analysis of a relatively outdated sample of Mastodon focusing on few aspects at a user level, while several open questions have not been answered yet, especially at the instance level. In this work, we aim at pushing forward our understanding of the Fediverse by leveraging the primary role of Mastodon therein. Our first contribution is the building of an up-to-date and highly representative dataset of Mastodon. Upon this new data, we have defined a network model over Mastodon instances and exploited it to investigate three major aspects: the structural features of the Mastodon network of instances from a macroscopic as well as a mesoscopic perspective, to unveil the distinguishing traits of the underlying federative mechanism; the backbone of the network, to discover the essential interrelations between the instances; and the growth of Mastodon, to understand how the shape of the instance network has evolved during the last few years, also when broading the scope to account for instances belonging to other platforms. Our extensive analysis of the above aspects has provided a number of findings that reveal distinguishing features of Mastodon and that can be used as a starting point for the discovery of all the DOSN Fediverse.

Large-scale national and transnational commercial land transactions, or Large-Scale Land Acquisitions (LSLAs), have been gaining a lot of academic attention since the late 2000s and since the reported rush for land, resulting in turn from an increase in demand for arable land. If many data exist to characterize land deals, the analysis of investment networks remain limited and predominantly portrays power asymmetries between countries from the Global North investing in the Global South. The aim of this work is to perform a deeper investigation on the land trade market, specifically focusing on cases that do not follow such narratives. For instance, almost 25% of the countries included in the transnational land trade network do not follow a strict investor/target dichotomy, thus being characterized by a double role, i.e., they both acquire and cede land in the transnational context. In order to globally acknowledge for what was currently considered as abnormal cases, we model open access data about LSLAs extracted from the Land Matrix Initiative (LMI) open-access database into a network graph, and adapt an eigenvector based centrality method originally conceived for online social networks, namely LurkerRank, to identify and rank anomalous profiles in the land trade market. We take into account three different network snapshots: a multi-sector network (including all the transnational deals in the LMI database), and three networks referring to specific investment sectors (agriculture,mines and biofuels). Experimental results show that emerging economies (e.g., China and Malaysia) play a central role in the land trade market, by creating alternative dynamics that escape the classic North/South one. Our analyses also show how African countries that are often seen as targets of land trade transactions in a specific sector, may often acquire foreign land in the context of investments in the same sector (i.e., Zimbabwe for biofuels and the Democratic Republic of Congo for the mining sector).

Abstract To properly capture the complexity of influence propagation phenomena in real-world contexts, such as those related to viral marketing and misinformation spread, information diffusion models should fulfill a number of requirements. These include accounting for several dynamic aspects in the propagation (e.g., latency, time horizon), dealing with multiple cascades of information that might occur competitively, accounting for the contingencies that lead a user to change her/his adoption of one or alternative information items, and leveraging trust/distrust in the users’ relationships and its effect of influence on the users’ decisions. To the best of our knowledge, no diffusion model unifying all of the above requirements has been developed so far. In this work, we address such a challenge and propose a novel class of diffusion models, inspired by the classic linear threshold model, which are designed to deal with trust-aware, non-competitive as well as competitive time-varying propagation scenarios. Our theoretical inspection of the proposed models unveils important findings on the relations with existing linear threshold models for which properties are known about whether monotonicity and submodularity hold for the corresponding activation function. We also propose strategies for the selection of the initial spreaders of the propagation process, for both non-competitive and competitive influence propagation tasks, whose goal is to mimic contexts of misinformation spread. Our extensive experimental evaluation, which was conducted on publicly available networks and included comparison with competing methods, provides evidence on the meaningfulness and uniqueness of our models.

Trust inference is essential in a plethora of data mining and machine learning applications. Unfortunately, conventional approaches to trust inference assume trust networks are available, while in practice they must be derived from social network features. This is however a difficult task which has to cope with challenges relating to scarcity, redundancy and noise in the available user interactions and other social network features. In this work, we introduce the new problem of Trust Network Inference (TNI), that is, inferring a trust network from a sequence of timestamped interaction networks. To solve the TNI problem, we propose a principled approach based on a preference learning paradigm, under a preference-based racing formulation. The proposed approach is suitable for addressing the above challenges, moreover it is versatile (i.e., independent from the social network platform) and flexible w.r.t. the use of topological and content-based information. Extensive experimental evaluation focusing on two distinct ground-truth scenarios, has provided evidence of the meaningfulness and uniqueness of our TNI approach, which can be regarded as key-enabling for any application that requires to handle a trust network associated with a social environment.

To understand airline transportation networks (ATN) systems we can effectively represent them as multilayer networks, where layers capture different airline companies, the nodes correspond to the airports and the edges to the routes between the airports. We focus our study on the importance of leveraging synthetic generative multilayer models to support the analysis of meaningful patterns in these routes, capturing an ATN’s evolution with an emphasis on measuring its resilience to random or targeted attacks and considering deliberate locations of airports. By resorting to the European ATN and the United States ATN as exemplary references, in this work, we provide a systematic analysis of major existing synthetic generation models for ATNs, specifically ANGEL, STARGEN and BINBALL. Besides a thorough study of the topological aspects of the ATNs created by the three models, our major contribution lays on an unprecedented investigation of their spectral characteristics based on Random Matrix Theory and on their resilience analysis based on both site and bond percolation approaches. Results have shown that ANGEL outperforms STARGEN and BINBALL to better capture the complexity of real-world ATNs by featuring the unique properties of building a multiplex ATN layer by layer and of replicating layers with point-to-point structures alongside hub-spoke formations.