The research, developed through the collaboration between the Sapienza Research Unit and the University of Turin, is structured around a dual objective. On the one hand, to analyze how and why readers intensify their practices of augmented reading, identifying recurring strategies and capturing their complexity through the construction of interpretive archetypes of “augmented readers.” On the other hand, to design and test a Retrieval-Augmented Generation architecture applied to a personal collection, assessing its accuracy, relevance, and ability to reconstruct connections, as a tool for mediation and documentary enhancement in the digital environment.
The Sapienza Unit developed a research design structured around four integrated lines of inquiry—university focus groups, a reading diary, participatory comparison workshops in upper secondary schools, and a mixed quantitative-qualitative questionnaire addressed to Generation Z youth—with the aim of conducting an in-depth analysis of reading and augmented reading practices. The integration of qualitative and quantitative tools made it possible to provide a nuanced picture of the modes, motivations, and tensions that currently shape the reading experience. The findings show that reading retains a strong symbolic and cultural value, but is at the same time perceived as a cognitively demanding activity, requiring time, concentration, and mental availability, thus competing with other forms of media consumption. In this scenario, the intensification of reading does not emerge as an anomaly or merely as an effect of distraction; rather, it appears as a widespread, situated, and consciously negotiated practice: a tool for enhancing understanding, continuing without losing the thread, and building connections and meaningful interpretive maps. The modes of augmentation that emerged from the research do not take the form of rigid, separate categories, but rather of combinable dimensions that are activated in different ways depending on contexts, goals, and readers’ subjective dispositions. What distinguishes the observed practices, therefore, is not the presence or absence of augmentation per se, but rather which modalities tend to prevail, in which situations, and with what perceived effects, ranging from enrichment to slowing down or distraction. An articulated space of augmented reading thus emerges, structured along two axes: on the one hand, the purpose of augmentation (from functional to exploratory), and on the other, its timing and intensity (from synchronous and rapid to asynchronous and prolonged). The resulting diagram identifies four main areas: Functional + Synchronous; Functional + Asynchronous; Exploratory + Synchronous; Exploratory + Asynchronous. This matrix does not serve a merely classificatory function, but an operational one: it constitutes an analytical tool for interpreting practices, recognizing their tensions, and designing targeted educational interventions, in line with the action-research approach that characterizes the BIBLAB laboratory. Starting from this combinatorial structure, a set of interpretive archetypes of augmented readers was outlined—hybrid-pragmatic reader; strategic reader; immersive-defensive reader; normative-performative reader; expansive-bulimic reader; cartographic-reflexive reader; curious-immersive reader. These are not rigid typologies or fixed psychological profiles, but recurring configurations of practices, motivations, and stances toward the augmentation of reading. The archetypes make it possible to understand how augmentation is, in turn, normalized, strategically managed, avoided, planned, intensified, or incorporated into a reflective practice. The same individual may move across multiple configurations at different moments, depending on the text, the context, and the material conditions of reading.
Within the framework of the research activities developed under the PRIN project, the University of Turin Unit designed and tested a Retrieval-Augmented Generation (RAG) architecture applied to a digital collection constructed as a closed and controlled corpus for experimental and evaluative purposes. The collection integrates bibliographic materials, archival documents, and open-access resources, selected and processed as explicitly identified primary and secondary sources, normalized, segmented, and enriched with descriptive metadata, in order to define a rigorous informational perimeter within which the system operates in an exclusive and transparent manner. The adopted architecture maintains a strong anchoring to the sources: starting from natural language queries, the system activates a retrieval phase based on semantic similarity and provides the language model exclusively with the relevant textual segments, thus avoiding the introduction of information external to the corpus. The output thus consists of responses generated through the reorganization, synthesis, and cross-referencing of content that is effectively present in the sources. The experimentation, conducted through a conversational interface and evaluated by means of a structured set of differentiated questions, investigated accuracy, relevance, and contextual reconstruction capacity, positioning RAG as a tool for advanced documentary mediation and as a replicable model for the enhancement of complex collections in the digital environment.