Semantic Technologies for Scientific Literature Management (v2)

点击翻译按钮获取中文摘要

A typical day in the life of a researcher involves finding, reading and analyzing scientific literature that are related to her task at hand. Online digital libraries, academic publishing databases and open access repositories grant access to a plethora of information, which poses major challenges to researchers when trying to obtain the information they need: Most modern search engines return thousands of articles in a matter of seconds and researchers have to manually comb through the results to find the information that may aid them in their research activities - a time-consuming and laborious task, during which critical knowledge can be easily missed. This apparent disproportion of humans' capabilities in processing information versus the unprecedented rate of scientific publishing growth has encouraged researchers from various domains to look for new approaches to extract, organize, and manage knowledge from the immense amount of available literature in ever-growing repositories. The frontiers of these explorations are researchers from the thriving domain of semantic publishing - a synergic community of semantic web researchers, computational linguists, librarians and publishing companies aiming towards a platform for dissemination of scientific literature accessible to both humans and machines. With this proposal, we are joining the movement, by developing new methodologies for transforming scientific literature from written content in isolation into a queryable web of documents suitable for machine interpretation. We conceptualize the myriad of possibilities that such a knowledge base can provide for researchers in form of a Personal Research Agent that helps researchers in finding, writing and learning from scientific literature. We demonstrate how a confluence of state-of-the-art techniques from the Semantic Web and Natural Language Processing domains can realize a proactive agent that can offer personalized services to researchers in retrieval and understanding of scientific literature, based on their background knowledge, interests and tasks. The agent's knowledge base, is populated with knowledge automatically extracted from scientific literature of a given domain using text mining techniques. We represent the structure and semantics of scientific literature to the agent by formalizing its rhetorical blocks and topics discussed within its content using semantic web technologies and the wealth of knowledge available in the Linked Open Data (LOD) cloud. Similarly, we introduce researchers to the agent through their \emph{profiles} that contain a formal representation of their personal metadata, interests, tasks and background knowledge. Our approach is the first to exploit the opportunities of linking a formal model of users and scientific publications in order to offer novel, personalized services to reduce a researcher's information overload.