Comparing CoreSC and AZ-II
A key issue in the project has been to evaluate the CoreSC annotation scheme against other schemas in order to assess their relative effectiveness. We compared CoreSC with the AZ-II annotation scheme on 36 chemistry papers. The two schemes are complementary in that they take different views on what a scientific paper represents. AZ assumes that a paper is the attempt of claiming ownership for a new piece of knowledge and aims to recover the rhetorical structure and the relevant stages in this argument. CoreSC on the other hand treats scientific papers as the humanly readable representations of scientific investigations. It therefore seeks to retrieve the structure of the investigation from the paper in the form of generic high-level Core Scientific Concepts. Thus, they have different focus with CoreSC containing more categories pertaining to the content of the paper whereas AZ categories elaborate on the path to various knowledge claims. The two schemes also differ in that CoreSC so far has used expert knowledge for annotation, whereas AZ-II has been annotated by expert-trained non-experts. The schemes have common ground, in the sense that they are both sentence based and target scientific papers. They even share some category names in common, such as “Method”, “Result” and “Conclusion”, even though these are defined differently in the two schemes and differ in granularity.
Our comparison showed a great deal of complementarity between the two schemes. However, CoreSC provides greater granularity when it comes to content-related categories, whereas AZ brings out the distinctive knowledge claims being made by in the paper being analysed. This suggests that a combined approach would be beneficial.
You can read the detailed analysis here [ lrecAZCoreSCfinal.pdf ]
Comments (0)