被引频次作为广泛使用的基础性学术评价指标，可以在一定程度上反映研究成果的学术影响力和研究主体的学术水平，然而这一指标也存在的天然的设计缺陷，由于它不考虑引用动机的复杂性和引用行为的多样性，因此无法全面反映学术影响力和学术水平的真相。为了深入挖掘引用的学术评价功能，本项目将学术评价的视角从被引端转向施引端、从被引频次转向引用行为、从唯量化评价转向量性结合，构建了一个基于引用行为的分类评价体系。首先，整合施引端的引用行为信息与被引端的引文特征信息，设计一个包含核心引用得分、正面引用得分、方法类引用得分、经典引用得分、权威引用得分和跨学科引用得分六个维度的引用行为指标体系。其次，依托大数据平台，通过海量结构化全文数据爬取和学术信息抽取，开展大规模的实证研究，验证这一评价体系的信度和效度。最后，仿照InCites和Altmetrics等网站架构开发一个支持可视化交互的评价结果查询和展示平台。

As a widely used and basic academic evaluation index, citation count indicators can reflect the academic influence of the research outputs and the academic level of the research institutions to a certain extent. However, this indicator also has natural design flaws as it does not consider complex citation motivation and the diversity of citation behaviors. Thus, it can not fully reflect the truth of academic influence and academic level. In order to dig deeper into the academic evaluation function of citations, the project expands the perspective of academic evaluation from cited side to citing side, from “cited count” to “citing behavior”, by combining quantitative and qualitative evaluation, to construct a qualitative and classification evaluation system based on citation behavior with reasonable normalization and weighting algorithm. By integrating the citing behavior information (from citing paper) and cited feature information (from cited reference), we will design a new evaluation system with multiple different dimensions, such as the score of core citing, positive citing, methodology score, authoritative citing, interdisciplinary citing, etc. In order to verify the reliability and validity of this evaluation system, the project team will rely on the big data platform to carry out large-scale empirical research by crawling massively structured full texts and extracting academic information. Finally, we will develop and launch an evaluation result retrieval system and display platform that supports visual interaction, modeled on the InCites or Altmetrics website architecture.