Network regression with predictive clustering trees

Network data describe entities represented by nodes, which may be connected with (related to) each other by edges. Many network datasets are characterized by a form of autocorrelation, where the value of a variable at a given node depends on the values of variables at the nodes it is connected with. This phenomenon is a direct violation of the assumption that data are independently and identically distributed. At the same time, it offers an unique opportunity to improve the performance of predictive models on network data, as inferences about one entity can be used to improve inferences about related entities. Regression inference in network data is a challenging task. While many approaches for network classification exist, there are very few approaches for network regression. In this paper, we propose a data mining algorithm, called NCLUS, that explicitly considers autocorrelation when building regression models from network data. The algorithm is based on the concept of predictive clustering trees (PCTs) that can be used for clustering, prediction and multi-target prediction, including multi-target regression and multi-target classification. We evaluate our approach on several real world problems of network regression, coming from the areas of social and spatial networks. Empirical results show that our algorithm performs better than PCTs learned by completely disregarding network information, as well as PCTs that are tailored for spatial data, but do not take autocorrelation into account, and a variety of other existing approaches.     

PRILJ: an efficient two-step method based on embedding and clustering for the identification of regularities in legal case judgments

In an era characterized by fast technological progress that introduces new unpredictable scenarios every day, working in the law field may appear very difficult, if not supported by the right tools. In this respect, some systems based on Artificial Intelligence methods have been proposed in the literature, to support several tasks in the legal sector. Following this line of research, in this paper we propose a novel method, called PRILJ, that identifies paragraph regularities in legal case judgments, to support legal experts during the redaction of legal documents. Methodologically, PRILJ adopts a two-step approach that first groups documents into clusters, according to their semantic content, and then identifies regularities in the paragraphs for each cluster. Embedding-based methods are adopted to properly represent documents and paragraphs into a semantic numerical feature space, and an Approximated Nearest Neighbor Search method is adopted to efficiently retrieve the most similar paragraphs with respect to the paragraphs of a document under preparation. Our extensive experimental evaluation, performed on a real-world dataset provided by EUR-Lex, proves the effectiveness and the efficiency of the proposed method. In particular, its ability of modeling different topics of legal documents, as well as of capturing the semantics of the textual content, appear very beneficial for the considered task, and make PRILJ very robust to the possible presence of noise in the data.

     

Bridging logic and kernel machines

We propose a general framework to incorporate first-order logic (FOL) clauses, that are thought of as an abstract and partial representation of the environment, into kernel machines that learn within a semi-supervised scheme. We rely on a multi-task learning scheme where each task is associated with a unary predicate defined on the feature space, while higher level abstract representations consist of FOL clauses made of those predicates. We re-use the kernel machine mathematical apparatus to solve the problem as primal optimization of a function composed of the loss on the supervised examples, the regularization term, and a penalty term deriving from forcing real-valued constraints deriving from the predicates. Unlike for classic kernel machines, however, depending on the logic clauses, the overall function to be optimized is not convex anymore. An important contribution is to show that while tackling the optimization by classic numerical schemes is likely to be hopeless, a stage-based learning scheme, in which we start learning the supervised examples until convergence is reached, and then continue by forcing the logic clauses is a viable direction to attack the problem. Some promising experimental results are given on artificial learning tasks and on the automatic tagging of bibtex entries to emphasize the comparison with plain kernel machines.     

A proactive and resilient seismic risk mitigation strategy for existing school buildings

Different methods and procedures have been developed to define prioritisation strategies of retrofit interventions aimed at reducing the seismic risk of school buildings on a large territorial scale. However, these approaches fail to demonstrate how risk analysis has been used successfully to quantitatively assess and select the optimal risk management decision. This article proposes innovative and useful metrics to measure the potential costs and benefits related to the prioritisation of retrofit intervention and the resilience of the analysed school system by directly integrating engineering, organisational, socio-economic and political aspects in the realm of seismic resilience assessment. Based on probabilistic risk assessments considering the new vulnerability of the prioritised school buildings, these measures could predict the expected economic and functional losses associated with a disastrous seismic event, as well as the possible post-disaster recovery of the system. In order to help decision-makers in selecting the optimal mitigation strategy with a multidisciplinary and multidimensional perspective, different political scenarios, the relative prioritisations of interventions and their intervention options are also defined. The proposed framework is demonstrated in a complex case study of 1,825 public schools in the Lima metropolitan area, Peru. Policymakers, planners and engineering professionals could benefit from results.     

Differential Metabotypes in Synovial Fibroblasts and Synovial Fluid in Hip Osteoarthritis Patients Support Inflammatory Responses

Changes in cellular metabolism have been implicated in mediating the activated fibroblast phenotype in a number of chronic inflammatory disorders, including pulmonary fibrosis, renal disease and rheumatoid arthritis. The aim of this study was therefore to characterise the metabolic profile of synovial joint fluid and synovial fibroblasts under both basal and inflammatory conditions in a cohort of obese and normal-weight hip OA patients. Furthermore, we sought to ascertain whether modulation of a metabolic pathway in OA synovial fibroblasts could alter their inflammatory activity. Synovium and synovial fluid was obtained from hip OA patients, who were either of normal-weight or obese and were undergoing elective joint replacement surgery. The synovial fluid metabolome was determined by 1H NMR spectroscopy. The metabolic profile of isolated synovial fibroblasts in vitro was characterised by lactate secretion, oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) using the Seahorse XF Analyser. The effects of a small molecule pharmacological inhibitor and siRNA targeted at glutaminase-1 (GLS1) were assessed to probe the role of glutamine metabolism in OA synovial fibroblast function. Obese OA patient synovial fluid (n = 5) exhibited a different metabotype, compared to normal-weight patient fluid (n = 6), with significantly increased levels of 1, 3-dimethylurate, N-Nitrosodimethylamine, succinate, tyrosine, pyruvate, glucose, glycine and lactate, and enrichment of the glutamine–glutamate metabolic pathway, which correlated with increasing adiposity. In vitro, isolated obese OA fibroblasts exhibited greater basal lactate secretion and aerobic glycolysis, and increased mitochondrial respiration when stimulated with pro-inflammatory cytokine TNFα, compared to fibroblasts from normal-weight patients. Inhibition of GLS1 attenuated the TNFα-induced expression and secretion of IL-6 in OA synovial fibroblasts. These findings suggest that altered cellular metabolism underpins the inflammatory phenotype of OA fibroblasts, and that targeted inhibition of glutamine–glutamate metabolism may provide a route to reducing the pathological effects of joint inflammation in OA patients who are obese.     

Improved Second Generation Iron Pincer Complexes for Effective Ester Hydrogenation

Hydrogenation of esters to alcohols with a well‐defined iron ^iPr2PNP pincer complex has been recently reported by us and other groups. We now introduce a novel and sterically less hindered ^Et2PNP congener that provides superior catalytic activity in the hydrogenation of various carboxylic acid esters and lactones compared to the known complex. Successful hydrogenation proceeds under relatively mild conditions (60 °C) with lower catalyst loadings.

     

Distribution of T-2 and HT-2 toxins in milling fractions of durum wheat

The effect of processing on mycotoxin content in milling fractions has been investigated in 10 samples of durum wheat contaminated with T-2 and HT-2 toxins at levels ranging from 97 to 5,954 μg/kg (sum of T-2 and HT-2 toxins). Either naturally contaminated samples or samples artificially inoculated with Fusarium sporotrichioides under field conditions were used. A method based on liquid chromatography-tandem mass spectrometry coupled with immunoaffinity column cleanup was validated in-house for the simultaneous analysis of both toxins in a variety of matrices, including uncleaned wheat, cleaned wheat, screenings, bran, red dog, fine middlings, and semolina. Mean recoveries from samples spiked with T-2 and HT-2 toxins at levels of 100 μg/kg ranged from 85 to 107%, with relative standard deviations (RSDs) lower than 14%. The milling process led to an increase of T-2 and HT-2 toxin contents up to 13- and 5-fold in screenings and bran, respectively, compared with occurrence in the uncleaned wheat; however, an overall reduction of T-2 and HT-2 toxins by 54% (RSD, 20%) and 89% (RSD, 3%) was observed in cleaned wheat and in semolina, respectively.     

Catalytic Synergism in a C60IL10TEMPO2 Hybrid in the Efficient Oxidation of Alcohols

A novel fullerene [5:1]hexakisadduct bearing two 2,2,6,6‐tetramethylpiperidine 1‐oxyl (TEMPO) radicals and ten 1‐propyl‐3‐methylimidazolium bromide moieties has been synthesized and characterized. Such an C₆₀IL₁₀TEMPO₂ hybrid has been successfully employed as a catalyst in the selective oxidation of a wide series of alcohols and is highly active at just 0.1 mol% loading. Moreover, it can be easily recovered by adsorption onto a multi‐layered covalently‐linked SILP phase (mlc‐SILP) through a “release and catch” approach and reused for up to 12 cycles without loss in efficiency. Interestingly, a catalytic synergistic effect of TEMPO and imidazolium bromide moieties combined in the same hybrid has been clearly shown.

     

Thiazolium‐Based Catalysts for the Etherification of Benzylic Alcohols under Solvent‐Free Conditions

Thiazolium and imidazolium hybrid materials were prepared by radical reactions between a mercaptopropyl‐modified SBA‐15 mesoporous silica and bis‐vinylthiazolium or bis‐vinylimidazolium dibromide salts. These hybrid materials were characterized by several techniques and were employed in the etherification reaction of 1‐phenylethanol. Solvent‐free conditions at 160 °C under different gas phases (oxygen, air, nitrogen and argon) were used. The thiazolium‐based material displayed excellent performances. Further studies were carried out using unsupported thiazolium salts, with or without a methyl group at the C‐2 position of the thiazolium moiety. These studies allowed us to propose a reaction mechanism. The supported thiazolium‐based material was successfully used in the etherification reaction of two other benzylic alcohols and also in seven consecutive cycles. This work represents the first use of thiazolium‐based compounds as catalysts for the etherification reaction of alcohols.