Science of retracted science: a citation analysis of the arts and humanities domain

In the scholarly publishing domain, a retraction is raised when a specific publication is considered erroneous by the venue in which it appeared after it was published. The aim of this work is uncovering new insights and learn new important information to help us understand the retraction phenomenon in the arts and humanities domain. Our investigation is based on a methodology defined using quantitative and qualitative measures derived from previous studies in the transdisciplinary research field of “science of science” (SciSci). The designed methodology takes into account a general case of retraction and applies a citation analysis based on five phases. Citations to retracted publications (before and after their retraction) are gathered and characterized with a set of attributes, including general metadata and information extracted from citing entities’ full text. The annotated characteristics are further considered for a statistical and a textual analysis (i.e., a topic modeling analysis). The contribution of this thesis is grounded by addressing the following research questions: (RQ1) How did scholarly research cite retracted humanities publications before and after their retraction? (RQ2) Did all the humanities areas behave similarly concerning the retraction phenomenon? (RQ3) What are the main differences and similarities in the retraction dynamics between the humanities domain and the STEM disciplines? RQ1 and RQ2 are addressed by tuning and applying the methodology on the analysis of the retracted publications in the humanities domain. RQ3 is addressed on two levels, i.e., considering and comparing: (L1) the outcomes of the past studies on the retraction in STEM, and (L2) the results obtained from an analysis of a retraction case in STEM using the defined methodology.

Understanding, defining and assessing food systems’ social and economic sustainability indicators: implementation of mixed methodologies for policy insights

Sustainability encompasses the presence of three dimensions that must coexist simultaneously, namely the environmental, social, and economic ones. The economic and social dimensions are gaining the spotlight in recent years, especially within food systems. To assess social and economic impacts, indicators and tools play a fundamental role in contributing to the achievements of sustainability targets, although few of them have deepen the focus on social and economic impacts. Moreover, in a framework of citizen science and bottom-up approach for improving food systems, citizen play a key role in defying their priorities in terms of social and economic interventions. This research expands the knowledge of social and economic sustainability indicators within the food systems for robust policy insights and interventions. This work accomplishes the following objectives: 1) to define social and economic indicators within the supply chain with a stakeholder perspective, 2) to test social and economic sustainability indicators for future food systems engaging young generations. The first objective was accomplished through the development of a systematic literature review of 34 social sustainability tools, based on five food supply chain stages, namely production, processing, wholesale, retail, and consumer considering farmers, workers, consumers, and society as stakeholders. The second objective was achieved by defining and testing new food systems social and economic sustainability indicators through youth engagement for informed and robust policy insights, to provide policymakers suggestions that would incorporate young generations ones. Future food systems scenarios were evaluated by youth through focus groups, whose results were analyzed through NVivo and then through a survey with a wider platform. Conclusion addressed the main areas of policy interventions in terms of social and economic aspects of sustainable food systems youth pointed out as in need of interventions, spanning from food labelling reporting sustainable origins to better access to online food services.

Meaningful insights: explainability techniques for black-box models on tabular data

Artificial Intelligence (AI) and Machine Learning (ML) are novel data analysis techniques providing very accurate prediction results. They are widely adopted in a variety of industries to improve efficiency and decision-making, but they are also being used to develop intelligent systems. Their success grounds upon complex mathematical models, whose decisions and rationale are usually difficult to comprehend for human users to the point of being dubbed as black-boxes. This is particularly relevant in sensitive and highly regulated domains. To mitigate and possibly solve this issue, the Explainable AI (XAI) field became prominent in recent years. XAI consists of models and techniques to enable understanding of the intricated patterns discovered by black-box models. In this thesis, we consider model-agnostic XAI techniques, which can be applied to Tabular data, with a particular focus on the Credit Scoring domain. Special attention is dedicated to the LIME framework, for which we propose several modifications to the vanilla algorithm, in particular: a pair of complementary Stability Indices that accurately measure LIME stability, and the OptiLIME policy which helps the practitioner finding the proper balance among explanations' stability and reliability. We subsequently put forward GLEAMS a model-agnostic surrogate interpretable model which requires to be trained only once, while providing both Local and Global explanations of the black-box model. GLEAMS produces feature attributions and what-if scenarios, from both dataset and model perspective. Eventually, we argue that synthetic data are an emerging trend in AI, being more and more used to train complex models instead of original data. To be able to explain the outcomes of such models, we must guarantee that synthetic data are reliable enough to be able to translate their explanations to real-world individuals. To this end we propose DAISYnt, a suite of tests to measure synthetic tabular data quality and privacy.