Abstract: Dynamic identification of industrial technology barriers is a strategic cornerstone for breaking the core technology blockades and building secure and controllable industrial chains. Relying on patent texts and control lists, this study proposes an industry technology barrier identification method based on semantic and evolutionary perspectives. First, the LDA and Word2Vec models are combined to extract core technical topics, enhancing the accuracy of topic recognition. Then, a three-level screening mechanism based on technology topic identification, technology competition mapping, technology control association is constructed to pinpoint key industrial technology barriers. Finally, the Dynamic Topic Model (DTM) is applied to depict the temporal evolution of barrier topics and predict future trends. Taking the field of lithography as a case study, empirical results validate the accuracy and robustness of the proposed method: the results are highly aligned with national strategic priorities, and its evolution trajectories strongly correspond with multi-dimensional factors such as industrial milestones, regional policy incentives, and has significant reference value for policy planning and research layout. This study breaks through the limitations of traditional static analysis by constructing a full-cycle research paradigm of "feature identification-mechanism analysis-trend prediction", providing data-driven decision support for overcoming technological bottlenecks in the lithography industry. The proposed methodological framework can be extended to strategic domains such as semiconductor equipment and biopharmaceuticals for technical competition and defense studies.

Key words: Industrial technology barriers, Dynamic Topic Model (DTM), Lithography, Semantics and evolution, Evolution trend prediction