Abstract: Using full-text analysis to explore the calculation method of finer-grained citation indexes is conducive to better reflect the real citation situation, and provide reference for the selection and improvement of citation indexes. This paper will discuss the acquisition and calculation of citation indexes based on full-text analysis through the steps of citation full-text data acquisition, citation context recognition, citation information recognition and citation index calculation. Then taking 57 papers recommended on F1000 as examples to calculate the citation indexes based on citation frequency, citation location and citation emotion, the citation value is further analyzed. The results show that academic papers are cited many times in the same citing literature, and the traditional calculation method of citation based on references list could not truly reflect the citation situation of academic papers. Most of the citations to the empirical papers are neutral ones without obvious emotional tendencies, and few are negative ones. More than half of the citations occur in the result part of the citing literature, and the higher the proportion of citation occurs in the result part, the more preferred the paper is by domain experts.

Key words: Full-text analysis, Citation index, Citation location, Citation emotion, Citation frequency, F1000