Claudio Conversano

Preference learning, or the analysis of preference rankings, is gaining more and more importance in various scientific disciplines. Preference learning methods allow predicting preferences on a set of alternatives. The ingredients are a pool of evaluators and a set of objects or items to be ranked in order of preference. The rank aggregation problem must be solved in order to aggregate preferences or rankings with the aim to find a consensus or collective decision. Branch-and-bound-like procedures are usable up to problems involving a relatively small number of objects, say less than 200. When the number of items becomes very large, the rank aggregation problem becomes increasingly difficult to approach so that it is universally recognized as an NP-hard problem. Several heuristic methods have been proposed to provide increasingly accurate solutions. These assume the Kemeny axiomatic approach that better deals with tied rankings. In this paper, we adopt a strategy based on Particle Swarm Optimization by adapting procedures born to solve optimization problems in continuous spaces to discrete combinatorial optimization problems. A simulation study shows the performance of the proposed algorithm in a controlled environment. A benchmarking complex data set and two real world data sets with large number of items are considered. As a result, the proposed algorithm provides significant savings in computational time and comparable accuracy with respect to other recent algorithms.

In this paper, we propose TAWE (title assignment with word embedding), a new method to automatically assign titles to topics inferred from sets of documents. This method combines the results obtained from the topic modeling performed with, e.g., latent Dirichlet allocation (LDA) or other suitable methods and the word embedding representation of words in a vector space. This representation preserves the meaning of the words while allowing to find the most suitable word that represents the topic. The procedure is twofold: first, a cleaned text is used to build the LDA model to infer a desirable number of latent topics; second, a reasonable number of words and their weights are extracted from each topic and represented in n-dimensional space using word embedding. Based on the selected weighted words, a centroid is computed, and the closest word is chosen as the title of the topic. To test the method, we used a collection of tweets about climate change downloaded from some of the main newspapers accounts on Twitter. Results showed that TAWE is a suitable method for automatically assigning a topic title.

Aims: Epidemiological research has shown relevant differences between sexes in clinical manifestations, severity, and progression of cardiovascular and metabolic disorders. To date, the mechanisms underlying these differences remain unknown. Given the rising incidence of such diseases, gender-specific research on established and emerging risk factors, such as dysfunction of glycaemic and/or lipid metabolism, of sex hormones and of gut microbiome, is of paramount importance. The relationships between sex hormones, gut microbiome, and host glycaemic and/or lipid metabolism are largely unknown even in the homoeostasis status. Yet this knowledge gap would be pivotal to pinpoint to key mechanisms that are likely to be disrupted in disease context. Methods and results: Here we present the Women4Health (W4H) cohort, a unique cohort comprising up to 300 healthy women followed up during a natural menstrual cycle, set up with the primary goal to investigate the combined role of sex hormones and gut microbiota variations in regulating host lipid and glucose metabolism during homoeostasis, using a multi-omics strategy. Additionally, the W4H cohort will take into consideration another ecosystem that is unique to women, the vaginal microbiome, investigating its interaction with gut microbiome and exploring—for the first time—its role in cardiometabolic disorders. Conclusion: The W4H cohort study lays a foundation for improving current knowledge of women-specific mechanisms in cardiometabolic regulation. It aspires to transform insights on host–microbiota interactions into prevention and therapeutic approaches for personalized health care.

Network-based Semi-Supervised Clustering (NeSSC) is a semi-supervised approach for clustering in the presence of an outcome variable. It uses a classification or regression model on resampled versions of the original data to produce a proximity matrix that indicates the magnitude of the similarity between pairs of observations measured with respect to the outcome. This matrix is transformed into a complex network on which a community detection algorithm is applied to search for underlying community structures which is a partition of the instances into highly homogeneous clusters to be evaluated in terms of the outcome. In this paper, we focus on the case the outcome variable to be used in NeSSC is numeric and propose an alternative selection criterion of the optimal partition based on a measure of overlapping between density curves as well as a penalization criterion which takes accounts for the number of clusters in a candidate partition. Next, we consider the performance of the proposed method for some artificial datasets and for 20 different real datasets and compare NeSSC with the other three popular methods of semi-supervised clustering with a numeric outcome. Results show that NeSSC with the overlapping criterion works particularly well when a reduced number of clusters are scattered localized.

Climate change represents one of the biggest challenges of our time. Newspapers might play an important role in raising awareness on this problem and its consequences. We collected all tweets posted by six UK and US newspapers in the last decade to assess whether 1) the space given to this topic has grown, 2) any breakpoint can be identified in the time series of tweets on climate change, and 3) any main topic can be identified in these tweets. Overall, the number of tweets posted on climate change increased for all newspapers during the last decade. Although a sharp decrease in 2020 was observed due to the pandemic, for most newspapers climate change coverage started to rise again in 2021. While different breakpoints were observed, for most newspapers 2019 was identified as a key year, which is plausible based on the coverage received by activities organized by the Fridays for Future movement. Finally, using different topic modeling approaches, we observed that, while unsupervised models partly capture relevant topics for climate change, such as the ones related to politics, consequences for health or pollution, semi-supervised models might be of help to reach higher informativeness of words assigned to the topics.

This paper introduces the Bradley–Terry regression trunk model, a novel probabilistic approach for the analysis of preference data expressed through paired comparison rankings. In some cases, it may be reasonable to assume that the preferences expressed by individuals depend on their characteristics. Within the framework of tree-based partitioning, we specify a tree-based model estimating the joint effects of subject-specific covariates over and above their main effects. We, therefore, combine a tree-based model and the log-linear Bradley-Terry model using the outcome of the comparisons as response variable. The proposed model provides a solution to discover interaction effects when no a-priori hypotheses are available. It produces a small tree, called trunk, that represents a fair compromise between a simple interpretation of the interaction effects and an easy to read partition of judges based on their characteristics and the preferences they have expressed. We present an application on a real dataset following two different approaches, and a simulation study to test the model’s performance. Simulations showed that the quality of the model performance increases when the number of rankings and objects increases. In addition, the performance is considerably amplified when the judges’ characteristics have a high impact on their choices.

The iterative Threshold-based Naïve Bayes (iTb-NB) classifier is introduced as a (simple) improved version of the previously introduced non-iterative Threshold-based Naïve Bayes (Tb-NB) classifier. iTb-NB starts from a Natural Language text-corpus and allows the user to quantify with a numeric value a sentiment (positive or negative) from a specific test. Differently from Tb-NB, iTb-NB is an algorithm aimed at estimating multiple threshold values that concur to refine Tb-NB’s decision rules when classifying a text into positive (negative) based on its content. Observations with sentiment scores close to the threshold are marked to be reclassified, hence a new decision rule is defined for them. Such “iterative” process improves the quality of predictions w.r.t. Tb-NB but keeping the possibility to utilize its results as the input of useful post-hoc analyses. The effectiveness of iTb-NB is evaluated analyzing hotel guests’ reviews from all hotels located in the Sardinia region and available on Booking.com. Furthermore, iTb-NB is compared with Tb-NB in terms of model accuracy, resistance to noise, and computational efficiency.

Non-destructive, fast, and accurate methods of dating are highly desirable for many heritage objects. Here, we present and critically evaluate the use of near-infrared (NIR) spectroscopic data combined with three supervised machine learning methods to predict the publication year of paper books dated between 1851 and 2000. These methods provide different accuracies; however, we demonstrate that the underlying processes refer to common spectral features. Regardless of the machine learning method used, the most informative wavelength ranges can be associated with C-H and O-H stretching first overtone, typical of the cellulose structure, and N-H stretching first overtone from amide/protein structures. We find that the expected influence of degradation on the accuracy of prediction is not meaningful. The variance-bias decomposition of the reducible error reveals some differences among the three machine learning methods. Our results show that two out of the three methods allow predictions of publication dates in the period 1851-2000 from NIR spectroscopic data with an unprecedented accuracy of up to 2 years, better than any other non-destructive method applied to a real heritage collection.