• DARIAH EU
• Other literature type close
• European Commission close
• ISTI Open Portal close

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.

If citation is a common practice for publications, it is relatively new for data especially in SSH. This paper will present the work carried out during the SSHOC project about data citation in general and more precisely how to make them actionable. International audience

Abstract. The remote visualization and navigation of 3D data directly inside the web browser is becoming a viable option, due to the recent efforts in standardizing the components for 3D rendering on the web platform. Nevertheless, handling complex models may be a challenge, especially when a more generic solution is needed to handle different cases. In particular, archeological and architectural models are usually hard to handle, since their navigation can be managed in several ways, and a completely free navigation may be misleading and not realistic. In this paper we present a solution for the remote navigation of these dataset in a WebGL component. The navigation has two possible modes: the ”bird’s eye” mode, where the user is able to see the model from above, and the ”first person” mode, where the user can move inside the structure. The two modalities are linked by a point of interest, that helps the user to control the navigation in an intuitive fashion. Since the terrain may not be flat, and the architecture may be complex, it’s necessary to handle these issues, possibly without implementing complex mesh-based collision mechanisms. Hence, a complete navigation is obtained by storing the height and collision information in an image, which provides a very simple source of data. Moreover, the same image-based approach can be used to store additional information that could enhance the navigation experience. The method has been tested in two complex test cases, showing that a simple yet powerful interaction can be obtained with limited pre-processing of data.

Abstract The solid geometry of archaeological deposits is fundamental to the interpretation of their chronological sequence. However, such stratigraphic sequences are generally viewed as static two-dimensional diagrammatic representations which are difficult to manipulate or to relate to real layers. The ADS 3D Viewer is a web-based resource for the management and analysis of archaeological data. The viewer was developed to take advantage of recent developments in web technology, namely the adoption of WebGL (Web Graphics Library) by current web browsers. The ADS 3D Viewer combines the potential of the 3D Heritage Online Presenter (3DHOP), a software package for the web-based visualization of 3D geometries, with the infrastructure of the Archaeology Data Service (ADS) repository, in the attempt to create a platform for the visualization and analysis of 3D data archived by the ADS. Two versions of the viewer have been developed to answer the needs of different users. The first version, the Object Level 3D Viewer , was implemented to extend the browsing capability of ADS project archives by enabling the visualization of single 3D models. The second version, the Stratigraphy 3D Viewer , is an extension which allows the exploration of a specific kind of aggregated data: the multiple layers of an archaeological stratigraphic sequence. This allows those unable to participate directly in the fieldwork to access, analyse and re-interpret the archaeological context remotely. This has the potential to transform the discipline, allowing inter-disciplinary, cross-border and ‘at-distance’ collaborative workflows, and enabling easier access to and analysis of archaeological data.

Abstract 3D Heritage Online Presenter (3DHOP) is a framework for the creation of advanced web-based visual presentations of high-resolution 3D content. 3DHOP has been designed to cope with the specific needs of the Cultural Heritage (CH) field. By using multiresolution encoding, it is able to efficiently stream high-resolution 3D models (such as the sampled models usually employed in CH applications); it provides a series of ready-to-use templates and examples tailored for the presentation of CH artifacts; it interconnects the 3D visualization with the rest of the webpage DOM, making it possible to create integrated presentations schemes (3D + multimedia). In its design and development, we paid particular attention to three factors: easiness of use, smooth learning curve and performances. Thanks to its modular nature and a declarative-like setup, it is easy to learn, configure, and customize at different levels, depending on the programming skills of the user. This allows people with different background to always obtain the required power and flexibility from the framework. 3DHOP is written in JavaScript and it is based on the SpiderGL library, which employs the WebGL subset of HTML5, implementing plugin-free 3D rendering on many web browsers. In this paper we present the capabilities and characteristics of the 3DHOP framework, using different examples based on concrete projects.

The SSHOC Multilingual Terminologies consist of a set of multilingual metadata concepts and an automatically extracted multilingual Data Stewardship terminology. They have been created in the context of the SSHOC project as case studies, during an activity of investigation on how Language Technologies can help in promoting and facilitating multilingualism in the Social Sciences and Humanities. The SSHOC Multilingual terminologies has been created to evaluate performances of state-of-the-art tools and to derive a set of recommendations as to how best apply them. The talk will present the main recommendations derived from the investigation and the workflow and tools adopted to publish the Terminologies as SKOS resources.