• DARIAH EU
• Publications close
• Research data close
• Article close
• IT close
• ZENODO close

Digital archives of memory institutions are typically concerned with the cataloguing of artefacts of artistic, historical, and cultural value. Recently, new forms of citizen participation in cultural heritage have emerged, producing a wealth of material spanning from visitors' experiential feedback on exhibitions and cultural artefacts, to digitally mediated interactions like the ones happening on social media platforms. Citizen curation is proposed in the context of the European project SPICE - Social Participation, Cohesion, and Inclusion through Cultural Engagement - as a methodology for producing, collecting, interpreting, and archiving people's responses to cultural objects, with the aim of favouring the emergence of multiple, sometimes conflicting viewpoints, and motivating users and memory institutions to reflect upon them. We argue that citizen curation urges to rethink the nature of computational infrastructures supporting data management of memory institutions, bringing novel challenges that include issues of distribution, authoritativeness, interdependence, privacy, and rights management. To approach these issues, we survey relevant literature towards a distributed, Linked Data infrastructure, with a focus on identifying the roles and requirements involved in such an infrastructure. We show how existing research can contribute significantly in facing the challenges raised by citizen curation, and discuss challenges and opportunities from the socio-technical standpoint.

Obtaining high-quality labelled data for training a classifier in a new application domain is often costly. Transfer Learning (a.k.a. “Inductive Transfer”) tries to alleviate these costs by transferring, to the “target” domain of interest, knowledge available from a different “source” domain. In transfer learning the lack of labelled information from the target domain is compensated by the availability at training time of a set of unlabelled examples from the target distribution. Transductive Transfer Learning denotes the transfer learning setting in which the only set of target documents that we are interested in classifying is known and available at training time. Although this definition is indeed in line with Vapnik’s original definition of “transduction”, current terminology in the field is confused. In this article, we discuss how the term “transduction” has been misused in the transfer learning literature, and propose a clarification consistent with the original characterization of this term given by Vapnik. We go on to observe that the above terminology misuse has brought about misleading experimental comparisons, with inductive transfer learning methods that have been incorrectly compared with transductive transfer learning methods. We then, give empirical evidence that the difference in performance between the inductive version and the transductive version of a transfer learning method can indeed be statistically significant (i.e., that knowing at training time the only data one needs to classify indeed gives an advantage). Our clarification allows a reassessment of the field, and of the relative merits of the major, state-of-the-art algorithms for transfer learning in text classification.

One of the emerging needs of the archaeological community is represented by the importance of availing of systems that allow to tackle new research questions, by querying diverse available resources. Usually, archaeological digital data is stored in non-standardised individual databases with a limited possibility of integration and a high level of fragmentation. The EU-funded project ARIADNE, has developed an e-infrastructure which enables the integration of archaeological datasets from various different institutions, integrating resource discovery metadata using controlled vocabularies, thesauri, gazetteers and ontology (CIDOC CRM). This paper presents the ARIADNE infrastructure, describing the activities undertaken by the project to achieve interoperability of archaeological resources at the dataset and item level. Moreover, the architecture of the ARIADNE Infrastructure and the Portal, with the different ways to search and access the resources are described.

This paper describes the activities carried out under the ARIADNE project to demonstrate the item-level integration process of archaeological archives through the use of semantic technologies. To this end, some ancient coin records, coming from the archives of important European archaeological institutions, were selected. The subset thus created, has been carefully analysed by means of specific tools to identify similar concepts and common metadata elements that could serve as the basis for integration. CIDOC CRM was chosen as the conceptual model for encoding the identified entities, while some important numismatic vocabularies have been employed to improve standardisation. The implementation phase has benefited from the use of advanced tools for mapping and conversion of the original information in a semantic form (RDF), the creation of a triple store to place the newly integrated data and the necessary interfaces for accessing and querying them.