Personalised support to examine context dependency between history of science events more

Personalised Support to Examine Context Dependency between History of Science Events Ilaria Corda, Vania Dimitrova, and Brandon Bennett School of Computing, University of Leeds, UK {ilaria,vania,brandon}@comp.leeds.ac.uk Abstract. Users access digital library content to fulfill task-specific searches. In this position paper, we argue that semantically driven approaches are needed to offer personalised access to historical collections. We consider a case study in the History of Science and illustrate how an ontology can be used to generate personalised, contextualised reference space in order to help users examine the context of scientific events. Key words: Personalisation, History of Science, Ontology 1 Introduction The remarkable growth of digital content leads to its wide availability to a large diversity of users. User access to digital resources is based on specific tasks, needs, and requirements, and is influenced by the users’ background and preferences. Hence, personalisation can play a crucial role in enhancing user access to digital content by reducing information overload and helping users find the most relevant resources. Current advances in personalised access to digital content deal mostly with museums, online exhibitions, and 3D digital dossiers of artwork installations[10][5]. The key issues in such applications are modelling the visitors’ interests and recommending items the users may wish to see [1]. Research has focused on nonintrusive ways for gathering users’ preferences by deriving behaviour patterns, which can be combined with explicit user ratings [10]. Another issue relevant to museum domains is to enhance the chance of post-visits and improve the users’ experience by offering resources based on their previous visits [3]. Recent research shows that semantically-enhanced approaches can improve personalisation in online museum collections, e.g. ontologies have been used to generate navigation paths based on user interests [2], improve the user modelling algorithms [12], or to recommend connected content based on geo-spatial links [7]. In museum collections, the users are mainly from the general public and personalisation is driven mostly by user preferences and interests. These approaches are insufficient for personalisation in digital libraries, where users belong to specialised categories and commonly have task-specific search needs [6]. Digital libraries are vehicles for gaining knowledge on a topic. Users searching through digital libraries can be researchers, students, or subject specialists, who come 2 Ilaria Corda, Vania Dimitrova, Brandon Bennett with different background and look for content to accomplish particular tasks. Therefore, we argue that effective approaches for personalisation in digital libraries should be task-based rather than interest-based, and should take into account the specific user categories and their goals when searching for content. In addition, the specificity of the domain is projected into the user tasks. For example, when searching through Chemistry or Biology digital collections, researchers may wish to compare all experiments related to a particular phenomenon, while in a History domain users primarily investigate the dependency between historical events [4]. We argue that an appropriate description of the domain, in the form of an ontology, should be exploited for providing task-driven personalisation in digital libraries. As advocated by [9], modelling time dependency is a crucial requirement in order to deal with more advanced demands in historical domains, such as uncertainty (missing or unknown information), subjectivity (multiple interpretations), and vagueness (unprecise temporal boundaries). Furthermore, representing time dependency between historical events appears a critical requirement underlying task-based trajectory planning in digital collections. This position paper points at the need for personalisation to help users find relevant resources in the History of Science. We proposes the use of an ontology to generate a contextualised reference space tailored to the user’s task and domain background, and illustrate how this space can be used to help users examine the surrounding context of History of Science events. 2 Digital content used in the History of Science The History of Science studies how people’s understanding of science is changing over time. It is considered a subfield of the History of Ideas which more broadly studies the manifestations and changes of human ideas over time. The History of Science is a unique blend of History, Philosophy, Science, and Sociology which does not correspond to the history of particular sciences or merely to a sum of those. Instead, while scientific disciplines themselves are characterized by a strong dimension of technicality, the History of Science primarily focuses on the impact of social and philosophical influences to which all scientific advancements are subjected [11]. History of Science content can be structured in primary and secondary 1 : – Primary content: textual materials - published works such as books, pamphlets, theses, letters, etc; unpublished writing or manuscripts such as notebooks, diaries, projects; interviews. – Primary content: non textual materials - scientific instruments, iconographic materials such as maps, photographs, drawings; objects studied by scientists. 1 de Andrade Martins, R.: International Databases on History of Science Sources, Strategies for the Developments of Databases, 2003 Available at: http://www.ifi.unicamp.br/ ghtc/sources/sources1.htm Personalised Support to Examine Context Dependency 3 – Secondary content - historiographic studies, biographies, contextual sources such as cultural and political information about a scientific period; interdisciplinary works or works in related fields (mainly from Philosophy of Science or Sociology of Science). With the recent advancement in digitisation, many of the above resources have been made electronically available. A computational representation of these materials is defined as digital tertiary sources of information 2 , which include library catalogues, databases, archives, digital collections, and repositories. Historians of Science are studying in general how science functions, the relationships between science and society, why and how a community accepts, rejects or refuses certain scientific results. For this, they are increasingly using digital tertiary sources, e.g. Sciper 3 (an electronic index collecting nineteenth-century general periodicals of Science, Technology and Medicine), Starry Messenger 4 (a History of Astronomy collection developed by the Whipple Museum which includes mainly web pages referring to additional recommended reading), Guide to the History of Science 5 (wide range of resources for professionals and scholars such as reference works, funding information, directory of journals, conferences announcements, museums, special collections), PhilSci Archive 6 (e-prints in Philosophy of Science). The existing History of Science collections provide general search functionalities, some of which exploit metadata. A common problem is that the search is fully dependent on the user’s expertise and the appropriateness of the key words specified by him/her. For instance, if the users searches for Kepler, he/she will only get resources directly related to this scientist, e.g. Kepler’s laws of planetary motion or the Rudolphine Tables. A wider context including relevant Scientific Revolution events, connected scientists, etc. will not be provided, unless the user explicitly asks for this, e.g. Kepler was Brahe’s assistant. Semantic support has not been explored to enhance search functionality, as well as the user’s tasks and general background have not been taken into account. Next, we will propose how an ontology-based approach can be used to provide task-driven support when examining the context of historical events. 3 Personalised Support to Examine Context Dependency Scientific events are in general situated occurrences. A most common task when researching History of Science content is to examine the context of a particular event. If we consider an event E, its context can include other events that happened in the same time as E, the scientists involved in E and in the surrounding 2 3 4 5 6 de Andrade Martins, R.: International Databases on History of Science Sources, Strategies for the Developments of Databases, 2003 Available at: http://www.ifi.unicamp.br/ ghtc/sources/sources1.htm Sciper Collection: http://www.sciper.leeds.ac.uk/index.htm Starry Messenger: http://www.hps.cam.ac.uk/starry/ Guide to the History of Science: http://www.hssonline.org/guide/ PhilSci Archive: http://philsci-archive.pitt.edu/ 4 Ilaria Corda, Vania Dimitrova, Brandon Bennett events, the reciprocal influences on the scientific achievements, the political and social situation at the time of E, the geographically linked events to E, etc. For example, assume that a user U examines the scientific event E=invention of the telescope by Galileo in 1609. U would need to find information about the Heliocentric System, which was claimed to be true and required strong proof of concept, for which the telescope was one of the main tools. Furthermore, U would need to examine related events leading to major discoveries in Optics around the time of E. Depending on the background of the user, he/she may be presented with a dynamically composed contexualised reference space to examine particular aspects of the context of E. For example, a student may be directed to the Starry Messenger (the major galilean treatise based on observations made through the telescope), while a researcher can be directed in addition to relevant interdisciplinary sources, e.g. Transactions of the Optical Society points at relevant accomplishments in Optics related to the invention of the telescope. Context dependency can be helpful for studying controversies - one of the most exciting aspects examined by the History of Science [8]. Controversies are contradicting views or ideas which have coexisted at the same time period because the scientific communities could not reach an agreement. For instance, imagine that a user U is examining the coexistence of two contradicting systems S1=Ptolematic-Aristotelian cosmology and S2=Copernican-Galilean cosmology. S1 and S2 coexisted together throughout the 17th century. Although there was strong evidence for the validity of S2, the official oppositions to the Heliocentrism by the Roman Church imposed S1. In this case, simply pointing at resources would not be beneficial for U , especially if U does not have much knowledge of the two systems. It will be helpful instead to first point to U helpful information from the contexts of S1 and S2, such as the major points where the two systems contradict and the supporters and opponents of each system. U can then be referred to relevant primary and secondary sources related to both S1 and S2. The above functionality would require the following components: – Access to digital collections with appropriate metadata linked to an ontology. Currently, we are considering the possibility to have access to Sciper which has been developed at the University of Leeds. – Domain ontology describing the History of Science events. It should enable reasoning about contextual information. We are using an ontology developed in our earlier research [4] which includes classes and relations about the domain. For example, the main relations (e.g. invent or work with) can be seen as a semantic network through which U will be taken. Those might support the user’s navigation trajectories by taking U to the appropriate contextual reference space based on his/her background and tasks. – User conceptual model. We assume that the representation of this model will follow closely the representation of the domain ontology, so that an appropriate mapping between both models can be done. The user’s conceptual model can be used to refine the context space of an event and to decide what content the user should be referred to. Personalised Support to Examine Context Dependency 5 4 Current State and Future Work We have developed a History of Science ontology that illustrates how to conceptualise and reason about a historical domain, focusing on representing and reasoning about temporal dependencies between scientific events [4]. This enables us to perform general queries to extract the context of a scientific event, combining what, who, where, and when queries. We are currently developing a web-based interface to show how the ontology can be used to automatically generate contextual information about an event that can be used to offer links to relevant resources. Our next step will be to consider how to represent the background of a user and to take it into account when refining the contextual reference space. References 1. Lora Aroyo, Rogier Brussee, Peter Gorgels, Natalia Stash, and Yiwen Wang. Personalized museum experience: The rijksmuseum use case. In Museums on the Web 2007, San Francisco, CA, 2007. 2. Charles Callaway and Tsvi Kuflik. Using a domain ontology to mediate between a user model and domain applications. In In Proc. of the 1st International Workshop on New Technologies for Personalized Information Access, 2005. 3. Trevor Collins, Paul Muholland, and Zdenek Zdrahal. Semantic browsing of digital collections. In In the Proc. of the 4th International Semantic Web Conference, (ISWC05), 2005. 4. Ilaria Corda. Ontology-based representation and reasoning about the history of science. Master’s thesis, School of Computing, University of Leeds, 2007. 5. Anton Eliens and Yiwen Wang. Rate, recommend, regret - an expert-based approach to the personalization of guided tours. In In the Proc. of Workshop on Personalization Enhanced Access to Cultural Heritage (PATCH07), 2007. 6. Yannis Ioannidis, WolfTilo Balke, Tiziana Catarci, Rosta Farzan, and Jacek Gwizdka. Do digital libraries require anything special from personalization? In 10th Delos Thematic Workshop (PersDL 2007), 2007. 7. Tomi Kauppinen, Jari Vaatainen, and Eero Hyvonen. Creating and using geospatial ontology time series in a semantic cultural heritage portal. In European Semantic Web Conference (ESWC), 2008. 8. Peter K. Machamer, Marcello Pera, and Aristides Baltas. Scientific Controversies: Philosophical and Historial Perspectives. Oxford University Press, 2000. 9. G. Nagypal, R. Deswarte, and J. Oosthoek. Applying the semantic web: The vicodi experience in creating visual contexualization for history. Literary and Linguistic Computing, 20(3):pages 327–349., 2005. 10. Lloyd Rutledge, Lora Aroyo, Rogier Brussee, Natalia Stash, Yiwen Wang, Henriette Cramer, Vanessa Evers, and Satyan Ramlal. Recommendation with semantics for cultural heritage. In In Proc. of Workshop on Personalised Access to Cultural Heritage (PATCH), 2007. 11. George Sarton. A Guide to the History of Science. Chronica Botanica Company, 1952. 12. Yiwen Wang, Lora Aroyo, Natalia Stash, and Lloyd Rutledge. Interactive user modeling for personalized access to museum collections: The rijksmuseum case study. In In Proc. of the 11th Int. Conf. on User Modeling (UM07), 2007.