Data Science and Social Research II

Frontmatter

Digital Methods and the Evolution of the Epistemology of Social Sciences

Abstract

After ten years that the debate on big data, computation and digital methods has been a contested epistemological terrain between some who were generally optimistic, and others who were generally critical, a large group of scholars, nowadays, supports an active commitment by social scientists to face the digital dimension of social inquiry. The progressive use of digital methods needs to be sustained by an abductive, intersubjective and plural epistemological framework that allows to profitably include big data and computation within the different paradigmatic traditions that coexist in our disciplines. In order to affirm this digital epistemology it is critical to adopt a methodological posture able to elaborate research designs with and against the digital, trying to exploit what digital techniques can give as added value, but going to test their reliability, alongside others techniques, including qualitative ones.

Enrica Amaturo, Biagio Aragona

Restricted Cumulative Correspondence Analysis

Abstract

In the context of the non-iterative procedures for performing a correspondence analysis with linear constraints, a new approach is proposed to impose linear constraints in analyzing a contingency table with one ordered set of categories. At the heart of the approach is the partition of the Taguchi’s statistic which has been introduced in the literature as simple alternative to Pearson’s index for contingency tables with an ordered categorical variable. It considers the cumulative frequency of cells in the contingency tables across the ordered variable. Linear constraints are then included directly in suitable matrices reflecting the most important components, overcoming also the problem of imposing linear constraints based on subjective decisions.

Pietro Amenta, Antonello D’Ambra, Luigi D’Ambra

Determining the Importance of Hotel Services by Using Transitivity Thresholds

Abstract

Customers’ preferences related to the quality, the change, and the progress of their expectations have turned the quality in an indispensable competitive factor for hotel enterprises. The hotels have to evaluate the customer satisfaction and to assign to each factor a weight, expressing its importance for their customers. The aim of this paper is to evaluate the importance of hotel services. Our analysis involves more than 300 customers that answered to a survey and it takes into account five criteria: Food, Cleanliness, Staff, Price/benefit, and Comfort. To derive the ranking of preferences we used pairwise comparisons. The main issue linked to pairwise comparisons is the consistency of judgements. Transitivity thresholds recently proposed in literature give meaningful information about the reliability of the preferences. Our study shows how the use of ordinal threshold may provide a ranking of services different from that obtained by applying traditional consistency Saaty thresholds.

Pietro Amenta, Antonio Lucadamo, Gabriella Marcarelli

Staging Cancer Through Text Mining of Pathology Records

Abstract

Valuable information is stored in a healthcare record system and over 40% of it is estimated to be unstructured in the form of free clinical text. A collection of pathology records is provided by the Veneto Cancer Registry: these medical records refer to cases of melanoma and contain free text, in particular, the diagnosis. The aim of this research is to extract from the free text the size of the primary tumour, the involvement of lymph nodes, the presence of metastasis, and the cancer stage of the tumour. This goal is achieved with text mining techniques based on a supervised statistical approach. Since the procedure of information extraction from a free text can be traced back to a statistical classification problem, we apply several machine learning models in order to extract the variables mentioned above from the text. A gold standard for these variables is available: the clinical records have already been assessed case-by-case by an expert. The most efficient of the estimated models is the gradient boosting. Despite the good performance of gradient boosting, the classification error is not low enough to allow this kind of text mining procedures to be used in a Cancer Registry as it is proposed.

Pietro Belloni, Giovanna Boccuzzo, Stefano Guzzinati, Irene Italiano, Carlo R. Rossi, Massimo Rugge, Manuel Zorzi

Predicting the Risk of Gambling Activities in Adolescence: A Case Study

Abstract

Adolescent gambling is internationally considered a serious public health concern, although this phenomenon is less explored than adult gambling. It is also well known that the early onset age of gambling is a risk factor for developing gambling problems in adulthood. This study examined 7818 adolescents enrolled in 16 public high schools in Lombardy, a Northwest Italy region, between March 2017 and April 2018 and it is part of a larger study aimed at investigating dysfunctional behaviours of adolescents, with the purpose of identifying the factors that increase the risk of vulnerability and the protection factors able to reduce the incidence of pathological phenomena. The objective of the present study was to investigate the susceptibility of adolescents in high school to develop gambling problems, explained by some individual and social factors, and by the association with the use of substances, such as alcohol and tobacco and other risk-taking behaviours. Various modelling methods were considered from an imbalanced learning perspective, as the prevalence rate of high school students in the sample with problem gambling is equal to 6.1%.

Laura Benedan, Gianna Serafina Monti

Municipal Managers in Italy: Skills, Training Requirements and Related Critical Aspects

Abstract

Public Administration in Italy has been experiencing noteworthy transformations necessary to meet citizens’ requirements. Public managers must apply such transformations and so their skills should be redefined. The aim of this article is to identify the skills that public managers should have for a good administration of the local community and to rank them in order of importance. A survey was administered to a sample of public managers operating in the Veneto region of Italy. Then skills and training requirements underlined by the same managers have been analysed. The findings indicate that a rich set of 26 skills is required. Good teamwork, proactive behaviour and authoritativeness are the most important skills for municipal managers, whereas specific and technical knowledge does not characterize the role of managers. This is particularly true in large municipalities, in which internal structures are complex and external relationships even more so.

Mario Bolzan, Giovanna Boccuzzo, Marco Marozzi

Attitudes Towards Immigrant Inclusion: A Look at the Spatial Disparities Across European Countries

Abstract

This paper aims at investigating individuals’ attitudes towards immigrants across European countries. In particular, we analyze how socio-demographic characteristics, altruism, and political preferences shape inclusive behavior towards immigrants. We use data from the European Value Study providing information at the individual level for 16 European countries, observed in 1999 and 2008. The results show relevant differences in individuals’ attitudes across countries and a general tendency to be less inclusive in the more recent period.

Riccardo Borgoni, Antonella Carcagnì, Alessandra Michelangeli, Federica Zaccagnini

A Bibliometric Study of the Global Research Activity in Sustainability and Its Dimensions

Abstract

The scientific production on “sustainability” has been increasing in recent years. To better understand and characterize this trend, a bibliometric study of international papers on this subject has been developed. A total of 3,994 articles from 1985 to 2018 have been selected and analyzed in order to discover the research trends in this field and the main dimensions and words related to the term “sustainability” that are most commonly employed in the scientific literature. The research has been conducted in the Web of Science from ISI Web of Knowledge database with the aim of identifying the major themes, authors, areas, types of documents and the sources, titles, years of publication and countries of these publications, as well as the main themes related to the topic “sustainability”.

Rosanna Cataldo, Maria Gabriella Grassia, Carlo Natale Lauro, Marina Marino, Viktoriia Voytsekhovska

Big Data Marketing: A Strategic Alliance

Abstract

The progress of technology and science goes along with the vast richness of the cultural and material existence of individuals, so that each person’s behavior at any moment generates large quantities of traceable information and data. Through the rise of the Internet, the quantity of data is geometrically increasing. It is then quite complex to manage it with traditional data systems. Today, Big Data touch every side of any economic activity, from public transportation, communications, bank securities, insurance, to government, health, education and other public utilities. Concomitantly with the rise of cloud computing, cloud applications, the multiplicity of mobile devices, and the maturity of e-commerce giants’ data marketing systems such as Amazon and Google, Big Data marketing is becoming increasingly focused and is being used by most companies. Following the development of digital growth, the aim of the marketing industry is to collect large amounts of assorted customer-related behavior sales data on which to develop their marketing strategies. Facing the era of Big Data, most firms have abandoned their traditional marketing strategies and have decided to opt for a powerful system of Big Data analysis so as better to identify their own customer target and consequently increase sales and profit. This chapter emphasizes the urge to observe the practical implications of an in-depth investigation on the strategic relationship between Big Data and marketing.

Federica Codignola

Data Processing in a Healthcare National System

(With the Analysis of the Italian HNS)

Abstract

In modern society and economy, a “personal data” can be considered as an “asset” with an own intrinsic value: since the increasing speed of technological evolution (added to the borderless context given by Globalisation) led the International Regulators to consider both how to guarantee the rights of individual natural person and the impact of Big-Data processing and management on the society and on economical markets, including in this range both the public and private scope. Nowadays we are assisting to a natural evolution from Big-Data to Smart-Data, especially in medical's and pharma's fields – due to a large treatment of sensitive data – in which it is fundamental to focus on the balancing between advantages and obligations, through the correct application of the “accountability principle” of the GDPR.Artificial Intelligence, Machine Learning, IoT, and Smart Data, due to their nature and customization, give an added value to what can be called “Health 4.0”, i.e. that mechanism of close collaboration between the operators of the integrated health and pharmaceutical system, through the interaction of information and data. Nevertheless, many legal and ethical aspects have not been exploited yet, still giving some uncertainties on possible future evolutions.

Manlio d’Agostino Panebianco, Anna Capoluongo

Smart Tourism System in Calabria

Abstract

We describe a Smart Tourism System called SMARTCAL, designed in the context of a R&D project aimed at supporting the tourism development in Calabria (a region in the South of Italy). The system is designed by considering Points and Events of Interest (PEOI) and their relationship with the local transport systems and infrastructures. A proactive tourist tour planner algorithm is proposed to generate optimized itineraries based on static and dynamic profiling of the users. Social media data are also taken into account for recommendations.

Annarita De Maio, Daniele Ferone, Elisabetta Fersini, Enza Messina, Francesco Santoro, Antonio Violi

Spatial Localization of Visitors Mobile Phones in a Sardinian Destinations’ Network

Abstract

In the act of placing and receiving calls, or sending text messages, a mobile phone reports its presence to the closest cell towers and unveils its geographical position. Mobile phone providers collect in their call data records the information about the clients’ spatial localization. The use of such data sets represents an enormous scientific opportunity to detect the structure of social networks. Quantifying and understanding network features may help to obtain deeper insight into applications of practical importance. For instance, knowing the number of visitors moving from one destination to another is an information that reveals valuable knowledge for regional tourism planning. With this aim, we investigate how visitors of Sardinia (Italy) move across the island, according to the localization tracks revealed by their cell phones. In order to study such mobility patterns, we employ aggregated call data records to construct a spatial network that we analyze with the aid of statistical tools. We also provide a comparison between the movement behaviors of national and international visitors.

Anna Maria Fiori, Ilaria Foroni

The Role of Open Data in Healthcare Research

Abstract

The generation and storage of data have dramatically increased worldwide in the last two decades. Computing and networking capabilities combined with openness enhance the potential impact of the accumulated data, offering society an opportunity to drive massive social, political, and economic change. Open data is a recent approach. In summary, open data can be freely used, shared, and built-on by anyone, anywhere, for any purpose. Though health open data are not regularly available, it is estimated that the value of a more effective use of data resources in the US health care sector alone could be worth USD 300 billion annually. To date, Open Government data count more than 10,000 datasets in Italy but only a few concern healthcare. In this chapter, we will try to clarify what open data are and, after having recalled the principles of Open Government, the attention will draw on open data in the health and pharmaceutical context, focusing on the state of the art in Italy and worldwide.

Carlotta Galeone, Rossella Bonzi, Paolo Mariani

Social Epidemiology: The Challenges and Opportunities of Worldwide Data Consortia

Abstract

Over the last few decades, social epidemiology has developed as a solid epidemiology branch, focusing on understanding how social experiences influence population health. At the same time, growing of collaborative and interdisciplinary research led to the proliferation of multi-institutional consortia, able to assess and quantify risk-disease associations of interest with a higher degree of accuracy, to explore subgroups of the population, and to investigate interactions between environmental, genetic, and socioeconomic factors. Increasing evidence shows that low Socioeconomic Position (SEP) is a strong determinant of morbidity and premature mortality from selected non-communicable diseases, including several cancers. Thus, an accurate quantification of the impact of SEP on cancer risk is of major importance to plan public health interventions for cancer incidence and socioeconomic disparities reduction. Large data consortia as the Stomach Cancer Pooling (StoP) Project and the International Head and Neck Cancer Epidemiology (INHANCE), in which the University of Milan is proactively involved, allowed investigators to address the effects of education and household income, the main SEP determinants, on gastric and head and neck cancer, respectively, confirming the existence of a strong association between low SEP and those major neoplasms.

Carlotta Galeone, Rossella Bonzi, Federica Turati, Claudio Pelucchi, Matteo Rota, Carlo La Vecchia

Identification of Opinion Makers on Twitter

Abstract

Twitter is a social platform that helps share ideas quickly and concisely. Although the network offers equal rights to post short texts, the attention these messages attract frequently depends on a user’s status in the real world. Thus the tweets of real life high-profile opinion makers will a priori have a higher probability of spurring the interest of society than the messages from the so-called grassroots. The paper elaborates on the developed classifier that detects automatically such opinion makers on Twitter. The approach exploits the Mixed Effect Random Forests method combined with the features engineered from the Twitter data. The accuracy and the sensitivity of the proposed technique outperform the results of the other machine learning classifiers on the out-of-sample data.

Svitlana Galeshchuk, Ju Qiu

Modelling Human Intelligence Using Mixed Model Approach

Abstract

In many psychometric studies, the observations may be often on longitudinal outcomes pertaining to General (G) and Specific (S) factors of human intelligence along with other covariates. Modelling human intelligence under Generalized Linear Mixed Model (GLMM) framework received the attention of psychologists in understanding the variables associated with the outcomes. In this paper, we formulate (i) a suitable GLMM model for count data of human intelligence factors and (ii) further examine the association between the outcome variables of Spearman’s G and S factors of human intelligence using joint longitudinal modelling along with other covariates based on school lunch intervention data.

Thanigaivasan Gokul, Mamandur Rangaswamy Srinivasan, Michele Gallo

An Analysis of the Impact of Requirements on Wages Within Sectors of the Tourism Industry

Abstract

The definitions of professional roles have changed quickly in the last few years due to several factors, such as the spread of new technologies. The recruitment process represents a way to evaluate the skills that a candidate needs to have in the workplace. This paper aims to evaluate the requirements for new hires in the tourism sector. In particular, we analysed the profiles of 1.526 workers recruited in 2017 by The Adecco Group in Italy. In the first phase, a conjoint analysis is performed to evaluate skills in the tourism sector, while in the second phase, a multinomial logistic regression is carried out to obtain more in-depth knowledge regarding the most selected (or preferred) profiles by employers, among those evaluated.

Paolo Mariani, Andrea Marletta, Lucio Masserini, Mariangela Zenga

Big Data and Economic Analysis: The Challenge of a Harmonized Database

Abstract

The real challenge that in the nowadays society needs to be scientifically faced is to accurately handle the enormous flow of information that in an IT world can be tremendously powerful to analyse the social and economic changes. The huge flow of data that private organizations and public administrations are storing in their databases is a precious and important source of information to complete the official statistics yielded by the National Statistics Institutes but not exempt from obstacles and issues that need to be solved. The dimension of private/public databases has to be considered in the Data Science scenario and involves that set of problems related to the so-called Big Data. This chapter provides a first scientific successful attempt to merge administrative databases and official statistical data in the field of research referred to the real estate economy that still suffers the consequences of the dearth of a complete and harmonized data warehouse.

Caterina Marini, Vittorio Nicolardi

ROC Curve in GAMLSS as Prediction Tool for Big Data

Abstract

During the latest years, Big Data appears as one of the most innovative and growing scientific areas of interest. In this field, finding reliable methods to make accurate predictions represents one of the most inspirational challenges. In the following paper, the use of ROC (Receiver Operating Characteristic) Curve, a binary tool, often used for medical tests, has been used to make predictions. In particular, the attention is focused on the implementation of the ROC Curve in GAMLSS (Generalized Additive Models for Location Scale and Shape), semi-parametric models suitable for huge and flexible datasets.

Andrea Marletta

Social Media in Disasters. Big Data Issues in Public Communication Field

Abstract

With the growth and the changing nature of the (big) data, the role of social sciences researchers has been enhanced, producing an emerging assemblage of tools and techniques for managing and making sense of such data. Furthermore, a web content analysis (WCA) approach could become the basis for the use of techniques that enhance the relational context in which the production of messages and texts puts itself. In light of these premises, our contribution aims to explore the way in which new research strategies of WCA—in particular the adoption of a mixed-methoda perspective that moves back and forth qualitative and quantitative approach—could be useful in the analysis of social media use and functions in the process of disasters implementation. As disaster social media framework includes users such as communities, governments, individuals, organizations, and media outlets, the use of a broader range of techniques in scientific study of disaster social media effects could facilitate the creation of disaster social media tools in the public communication field.

Francesco Marrazzo, Gabriella Punziano

Divorce in Italy: A Textual Analysis of Cassation Judgment

Abstract

The dissolution of marriage is a complex social phenomenon that needs new topics of investigation, especially concerning the role of legal institutions in the conflict between partners. The research aims to identify the main issues that emerge in the institutional dimension of the phenomenon, identifying evolution and complexity of this within the sentences of the Italian Court of Cassation. Through judgments’ analysis we can trace the variety of the phenomenon and identify interpretations of law in line with the evolution of contemporary institutions. The sentences are inserted in a demographic framework and are subsequently explored with topic probabilistic model (Latent Dirichlet Allocation), aimed to trace latent topic. In conclusion, the topics extracted refer to three main-dimensions, one to the procedural phases, another concerns the difficulty of leaving the separation phase and ending up in divorce, and finally the debate on the social-economic measures of divorce maintenance.

Rosanna Cataldo, Maria Gabriella Grassia, Marina Marino, Rocco Mazza, Vincenzo Pastena, Emma Zavarrone

A Bayesian Mixture Model for Ecotoxicological Risk Assessment

Abstract

In ecotoxicological risk assessment, the estimation of a Species Sensitivity Distribution (SSD) is a routine method used to derive hazardous levels of concentrations for chemical substances. Here, we propose a Bayesian hierarchical approach leading to the definition of a new SSD. Our approach allows to use all information available at chemical-class-species levels to make inferential decisions. We estimate parameters via computer-intensive methods based on Markov Chain Monte Carlo methods, and we propose a way to derive the estimates of concern levels of toxicants that could be easily adopted in ecotoxicological risk management.

Sonia Migliorati, Gianna Serafina Monti

Virtual Encounter Simulations: A New Methodology for Generating Conflict Data

Abstract

This article presents a new methodology for generating knowledge about group conflicts by the use of social surveys, which mainly inform about characteristics and values of the interviewees. The method is based on the idea of simulating virtual encounters between pairs of persons from the same and different groups in order to determine the value-conflicts between the related individuals. For a more subjective assessment of the situation, inter-group conflicts are compared with intra-group conflicts. This results in a new typology, which allows to conceptualize asymmetrical conflict. The proposed method is applied to the analysis of the national identities in the French- and German-speaking parts of Switzerland. It turns out, that the two groups have less conflict about “Swissness” than the traditional methods of analysis suggest.

Georg P. Mueller

Is Public Service Motivation–Performance Relationship Mediated by Other Factors?

Abstract

Although the association between public service motivation (PSM) of public employees and Performance has received increased attention, there are yet inconsistencies in the literature regarding how the PSM–performance relationship may be mediated by other factors. This study is based on a sample of 618 Italian public teachers and considers a set of hypotheses in public education in which the relationship is mediated by person–organization fit (P-O fit) and organizational commitment (OC). This mediated relationship varies depending on how performance is considered.

Raffaela Palma, Anna Crisci, Luigi D’Ambra

A Classification Algorithm to Recognize Fake News Websites

Abstract

“Fake news” is information that generally spreads on the web, which mimics the form of reliable news media content. In this paper, we use a classifier to distinguish a reliable source from a fake news website based on information potentially available on websites, such as the presence of a “contact us” section or a secured connection. This framework offers a concrete solution to attribute a “reliability score” to news websites, defined as the probability that a source is reliable or not, and based on this probability a user can decide if the news is worth sharing.

Giuseppe Pernagallo, Benedetto Torrisi, Davide Bennato

A Comparative Analysis of the University Student Mobility Flows Among European Countries

Abstract

Higher education institutions’ policies aimed at increasing the number of credits gained by university students abroad. Thus, the analysis of the internationalization process and the factors pulling and pushing students in a foreign country to complete their higher education are important features for academic institutions. In line with some previous studies, the present contribution aims to analyse the trend of the Erasmus student mobility flows and to capture the role played by each country by using the social network analysis approach. Data on Erasmus student exchanges among countries are gathered from the European Union Open Data Portal and used to build the network between countries. The main findings suggest that some countries in Europe are more attractive in terms of the number of incoming and outgoing students.

Marialuisa Restaino, Ilaria Primerano, Maria Prosperina Vitale

A Preference Index Design for Big Data

Abstract

TripAdvisor is a business service that works as a reputation system to guarantee quality in tourism experience. This kind of new service is based on Big Data technologies and characterized by generating, managing and summarizing, even with rating indexes, a quantitative experimental size of information, representing a frontier issue for data analysis. These data are organized and offered to users by a filter system aimed at recommending consumer’s choices. Through a methodological design oriented to reward competitive quality, this acts as a crowdsourced evaluation system. In this paper, we suppose that information provided through the website can be biased because past reviews and ratings can affect the process of data production. On the basis of an empirical research for approximately 26.000 scores on TripAdvisor multipoint scale organized into 8-years time series and harvested by R software, we observe non-linear dynamics and skewed distribution among values of the scale. In our study, we observed that the main goal of crowd rating platforms is to extensively rank subsets of a population of units. This is achieved through the systematic employment of estimation techniques of evaluative measures. We propose a design of rating indexes that reflects the original missions of crowd rating: to pragmatically decrease the risk of a bad experience for the customer, to coherently benchmark, and to reliably rank a list of competing units.

Venera Tomaselli, Giulio Giacomo Cantone

Construction of an Immigrant Integration Composite Indicator through the Partial Least Squares Structural Equation Model K-Means

Abstract

Integration is a multidimensional process, which can take place in different ways and at different times in relation to each of the single economic, social, cultural, and political dimensions. Hence, examining every single dimension is important as well as building composite indexes simultaneously inclusive of all dimensions in order to obtain a complete description of a complex phenomenon and to convey a coherent set of information. In this paper, we aim at building an immigrant integration composite indicator (IICI), able to measure the different aspects related to integration such as employment, education, social inclusion, active citizenship, and on the basis of which to simultaneously classify territorial areas such as European regions. For this application, the data collected in 274 European regions from the European Social Survey (ESS), Round 8, on immigration have been used.

Venera Tomaselli, Mario Fordellone, Maurizio Vichi

Facebook Debate on Sea Watch 3 Case: Detecting Offensive Language Through Automatic Topic Mining Techniques

Abstract

Over the years, there has been growing concern about the disproportionate use of hate speech on social media platforms. In this paper, we present a text analysis for detecting abusive language in Italian messages on Facebook, surrounding the debate over the migrant-rescue ship, Sea Watch 3, and its captain Carola Rackete. The study data consists of more than 130,000 posts retrieved from two pages relating to Matteo Salvini, the leader of the Italian Lega political party, and from the official Facebook pages of five Italian newspapers. To explore the presence of offensive and hatred expressions in the corpus and to establish to what extent social users’ language differs, depending on the type of Facebook pages analysed, we ran a topic model based on Latent Dirichlet Allocation. We have complemented this approach with tools from semantic network analysis.

Alice Tontodimamma, Emiliano del Gobbo, Vanessa Russo, Annalina Sarra, Lara Fontanella

Martini’s Index and Total Factor Productivity Calculation

Abstract

The axiomatic property of consistency in aggregation is fundamental for example in the analysis of economic systems divided into sectors and sub-sectors, to ensure consistency of results at different levels of sectoral aggregation. The expenditure ratios index numbers satisfy this property by construction; however, they often do not satisfy other properties such as the factor reversibility. Martini’s index satisfies both properties and in this work it is applied to the problem of calculating the Total Factor Productivity (TFP), which typically refers to a context in which consistency in aggregation is important. In the proposed applications the results obtained are very encouraging and stimulate interest in further study.

Biancamaria Zavanella, Daniele Pirotta

Backmatter