Higher-level bibliographic services

Bibliographic services, including Web of Science/Knowledge, Scopus, CiteSeerX, zbMATH, Google Scholar, DBLP, MathSciNet, COBISS, arXiv and others, provide data about scientific works (papers, books, reports, etc.). They are usually used by individual users for searching publications on selected topics, and by institutions for research evaluation and planning. They are used also in data analysis for bibliometric and scientometric research. For this purpose the data on a selected topic are often transformed in the collection of bibliographic networks linking different entities (modes: works, authors, editors, journals, keywords, institutions, countries, languages, etc.). According to the IFLA Library Reference Model, the bibliographic universe is also represented as a multi-mode multi-relational network, enabling clustering on different levels, for example versions of papers (preprints, published) or editions of a monograph. The data collected in different bibliographic databases can be used to provide higher order bibliographic and bibliometric services (such as: what to read (contact/visit)? where to publish? reviewer selection; a candidate’s activity report draft for a career application; etc.) for different types of users (students, researchers, teachers, decision makers, journal editors, funding agencies, research institutions, database managers, etc.). The main goal of the project is the identification of potential higher order services and development of some prototype solutions. To support this goal we have to provide high quality data often obtained by combining data from different databases. We also have to develop new algorithms for some analytical problems. A crucial step in construction of bibliographic networks is the entity (works, authors, etc.) identification / resolution (resolving synonymy / homonymy of entity names / labels). It is a necessary step in combining data from different sources. One of the goals of the project is to start developing robust and high precision methods for entity resolution for specific types of entities. An important tool for analysis of collections of networks are the derived networks obtained by combining network normalization (fractional approach), multiplication of networks, and temporal networks based on temporal quantities. We intend to explore the possibilities offered by these approaches. With big datasets the speed and efficiency of execution is crucial. Hence the core algorithms have to be implemented in a fast C-like programming language which is also easy to use and easy to maintain. The obvious candidate is the Julia programming language which is steadily gaining its reputation as a fast data analytics language. Our goal is to start developing an open source Julia library for bibliographic network analysis. We believe that our methods, algorithms, tools and the derived high-quality data will provide the foundation of higher level services for different kinds of users. In summary, the objectives of the project include: * Identification of potential higher level services and development of several prototype bibliographic analyses and related tools, motivated by needs of selected end-users. * Development of methods and algorithms for high quality bibliographic entity resolution based on bibliographic network analysis. This will enable us to implement processing pipelines that can be applied on bibliographic databases in order to obtain periodically refreshed high quality and up to date bibliographic data. * Further development of methodologies and algorithms for analysis of bibliographic networks, based on our past research (2-mode networks, fractional approach, temporal networks and temporal quantities) motivated by specific types of analyses with emphasis on how the science is developing in “real-time” (based on preprints).