Full-Text Citation Analysis And Applications

With the appearance of full-text literature repositories, Citation Analysis based on the large sample becomes increasingly feasible. Full-text citation analysis, by identifying the location, intensity and context of each citation instance in citing papers, could be used to characterize patterns and properties of citation behavior, uncover motivations and mechanisms behind the behavior of makingcitations, and reveal the roles and functions of the references.On the basis of previous studies of citation behavior, we propose a systematic and comprehensive framework for full-text Citation Analysis, which is built on three primary dimensions, namely, citation location, citation intensity, and citation context. In this dissertation, we selected Journal of Informetrics (JOI) as a case study. All the papers of JOI in XML format were downloaded from Elsevier(?) ConSyn and imported into the self-developed system for extracting citations. In the above three dimensions, the extracted citations were studied in detail.In the dimension of citation location, we find an uneven distribution of citation locations. Nearly half of the citations are located in the first30%of a paper, which essentially corresponds to the section of "Introduction." Citations usually occur in clusters for enhancing arguments of cited references. In general, the earlier a paper was published, the nearer to the beginning of the paper its citations are likely to appear. Highly cited references are often cited before citations to other references.In the dimension of citation intensity, we find that each reference is cited about1.5times in one citing article on average. It means it is quite common for a reference being cited more than once in one article. Multiplecitations of the same reference are more likely to appear in the same section than across different sections. There is no significant correlation between citation intensity and cited times of a reference. However, recently published references have higher citation intensity than old ones.In the dimension of citation context, we find that citation context in different sections or at different levels can be quite distinct in terms of content phrases and cue phrases. For example, citation contexts in the section of "introduction" are about research background; while in subsequent sections, citation contexts are about methodology or results. For multiple citations, citation context of the first citation has a higher level of abstraction, while subsequent citation contexts are more granual. The age and cited times of a reference is also related with its citation context. For example, most citation contexts of highly cited references take a third-person perspective, while those of less frequently cited references are in first-person perspective.Full-text Citation Analysis has important theoretical and practical implications. It promotes the development of bibliometrics, expand the scope and depth of classical citation analysis and enhance our understanding of citation behavior. Finally, we employed three case studies to demonstrate the potential of applying the new Citation Analysis in the fields of scientific literature retrieval, scientific evaluation and forecast, and science mapping.