【Purpose/significance】 In the era of big data, due to the lack of targeted and practical policies, the issue of protecting per⁃

sonal privacy information has not been effectively addressed. Therefore, this article studies the development process, disciplinary struc⁃ture, and past hotspots of research directions in this field from the perspective of biblioinformatics, based on a collection of research lit⁃erature.【Method/process】 This article takes research literature in the field of legal protection of personal privacy information as an ex⁃ample. First, a new potential feature topic model, Improved LDA, is proposed and compared with the traditional LDA model. Secondly,the improved LDA model is used to identify research topics in the field of legal protection of personal privacy information. Finally,time series is introduced to explore the temporal evolution characteristics of each research topic, in order to reveal the development sta⁃tus of the field in the past and predict its future development trend.【

Result/conclusion】 Research has found that the Improved LDA model performs significantly better than traditional LDA models in both topic consistency calculation and topic clustering. At the same time, an improved LDA model was used to discover eight thematic categories, including "research and development of information pro⁃tection technology, user information security in e-commerce platforms, personal privacy information security in the financial field, and personal privacy information security in the context of the epidemic".【Innovation/limitation】 This article takes research literature in the field of legal protection of personal privacy information as an example, and explores the knowledge structure, research status, and evolution laws of this field from the perspective of biblioinformatics. This provides effective reference for guiding the development of this research field, with the aim of providing reference for the optimization of existing policies and regulations. In addition, the univer⁃sality of the improved LDA model proposed in this article still needs to be tested.