[Purpose/significance] This study analyzed the Personalized Recommendation Service(PRS) of the Online Academic Resource Platform(OARP), provides a realistic basis for the in-depth study of personalized recommendation from the user's perspective, and provides a reference for identifying and bridging the research gap between academia and industry. [Method/process] The content analysis is used to conduct two rounds of coding analysis on the PRS provided by 23 OARP, and the display form, recommended content, recommendation explanation, recommendation time node, and recommendation type are selected as analysis indicators. [Result/conclusion] ①The display forms of person? alized recommendation on the OARP can be divided into highlighting, non-differentiated, navigation index, pop-up, prompt web page jump, and highlighting are used more; ②Recommended content is hyperlinks of scientific literature and its attributes; ③Recommended explanations on different platforms are similar, but the level of detail is different; ④The recommended time node mostly occurs after the user submits the search term, during to browsing the document details page, during the downloading; ⑤88.89% of the recommendation types provided by the platform are static recommendations. The failure of the personalized recommendation algorithm to incorporate the user's current interaction behavior, the demand for personalized recommendation, and the contextual factors into the recommendation algorithm is an important reason why users cannot actually feel the PRS with excellent performance.