Abstract: 【Purpose/significance】In order to reflect the core connotation intuitively andminingthe semantics of the policy text, this pa?
per focuses on topic mining,and carries out a structuredparseof the policy text from the perspectives of external attributes and content
attributes, which willprovide a new mode for the interpretation of policy text.【Method/process】LDA2Vec topic model is used in this
study to recognizecontext-based policytopic, as well as external attributes are extracted based on location and grammatical rules.There? by, we constructa descriptive framework for structuredparse of policy text.【Result/conclusion】In the empirical analysis, this frame? workwas successfully applied to interpret the "Internet +" policy text, indicating that the framework this paper proposed is help for dis? play policy elements visually, reveal the distribution of policy topics effectively, and perform large-scale batch analysis and interpreta? tion of policy textsefficiently.【Innovation/limitation】According to the structural parse framework, the formal and thematic features of the policy text are displayed, helping policy-related groups grasp the characteristics and focus of policy. At present, the in-depth con? tent interpretation needs further research.