2026 Vol.9 Apr.N02

• Title: Development of a Deep Learning Model for Evaluating the Value of Patent Commercialization in Universities from the Perspective o
• Name: Qinghua Fan， Jian Shen
• Company: Jilin Communications Polytechnic, Changchun 130012, China
• Abstract:

   Most methods for evaluating the commercialization value of university patents rely on 

  expert scoring or simple statistical analysis, calculating a total score by setting a series of indicators and applying their respective weights. However, due to the complex relationships and redundancy among these indicators, the evaluation accuracy is often poor. To address this, this paper proposes a method for constructing a deep learning model to evaluate the commercialization value of university patents from the perspective of new-quality productive forces.We introduce the historical patent commercialization record of the first inventor as a validation metric for the anchoring effect, while comprehensively considering both the basic characteristics of the patent and the characteristics of the patent inventor to construct a quantitative system for patent commercialization value. By calculating the mean and standard deviation of each indicator and using the correlation matrix to compute the variance contribution rate and cumulative variance contribution rate, we achieve dimensionality reduction of the quantitative indicators for university patent commercialization value.A Backpropagation (BP) neural network model is constructed. The dimension-reduced quantitative indicator data for university patent commercialization value is fed into the model for forward propagation. By applying activation functions to calculate the outputs of each layer, predicted values are derived, thereby achieving a quantitative assessment of university patent commercialization value.In the experiment, the accuracy of the proposed method was evaluated. The test comparison results clearly show that when using the proposed method to evaluate the commercialization value of university patents, the average agreement between the predicted commercialization value and the actual commercialization value is 94.6%, demonstrating a relatively ideal evaluation effect.

  
  

• Keyword: perspective of new-quality productive forces; university patents; commercialization value; evaluation methods; deep learning models;
• DOI: 10.12250/jpciams2026090401
• Citation form: Qinghua Fan,Jian Shen.Development of a Deep Learning Model for Evaluating the Value of Patent Commercialization in Universities from the Perspective of New-Quality Productive Forces[J]. Computer Informatization and Mechanical System,2026,Vol.9,pp.

[1] Yang Y. Rapid Patent Quality Evaluation Method Based on Big Data Analysis: Chinese Invention Patents as a Sample [J]. Advances in Economics, Management and Political Sciences, 2023, 7(1):343-349.

[2] Ren Z, Mei S, Deng Y. Impact of University-Inventor Patent Co-Ownership Reform on University Patenting Activities [J]. Journal of Nonlinear and Convex Analysis, 2023, 24(7):1581-1594.

[3] Nguyen P Q, Mai T N. Legal Rules on Patent Exploitation and Management for Innovation in Vietnamese Universities [J]. Edelweiss Applied Science and Technology, 2024, 8(6):9652-9669.

[4] Anagnosti A, Drivas K. The Relation Between University Science Base and Follow-Up Patent Activity [J]. Journal of the Knowledge Economy, 2025, 16(3):12171-12192. 

[5] Chen Y, Zhu P, Ma J, et al. A Trust-Enhanced Patent Recommendation Approach to University-Industry Technology Transfer [J]. ACM SIGMIS Database: the DATABASE for Advances in Information Systems, 2024, 55(1):35-55. 

[6] Researchers Submit Patent Application, 'Method and System for Evaluation and Development of Automated Driving System Features', for Approval (USPTO 20230091986)[J]. Electronics Newsweekly, 2023(Apr. 11):6016–6020.

[7] Shimmura K, Kano S. Effects of Japan's past industry-academia collaboration reforms on university patent licensing income: the case of the University of Tokyo [J]. International Journal of Technology Transfer and Commercialization, 2023, 20(4):429-450. 

[8] Liu Y, Guo C, Zhang J, et al. [Improvement of high-quality evaluation criteria for Chinese patent medicines based on whole-process control]. [J]. China Journal of Chinese Materia Medica, 2023, 48(6): 1700–1704. 

[9] Zhang H. Construction of a Research Model for Overseas Dissemination of Chinese Literature Based on Deep Learning Models [J]. International Journal of Distributed Systems and Technologies, 2025, 16(1):1-17. 

[10] Limengyang L M, Mustafa Z, Shufang L. Research on the Construction of an Outcome-Based Education Platform Based on Deep Learning [J]. Edelweiss Applied Science and Technology, 2025, 9(2):1168-1179. 

[11] Kim G, Yang J. Performance Evaluation of Large Language Models for IPC Code Classification in Patent Documents [J]. The Journal of Korea Institute of Information, Electronics, and Communication Technology, 2025, 18(2):135-142.

[12] Yin Y. Big Data Analysis and Modeling Method of College Student Employment Management Based on Deep Learning Model [J]. International Journal of Web-Based Learning and Teaching Technologies (IJWLTT), 2023, 18(18):1-15. 

[13] Meshram A, Patil A, Mulik S, et al. Text Generation Model Using Deep Learning [J]. International Journal for Research in Applied Science and Engineering Technology, 2025, 13(1):23-25. 

[14] Balakumar N. Advanced Transfer Learning Strategies for Improving Deep Learning Model Efficiency and Accuracy [J]. International Journal for Research in Applied Science and Engineering Technology, 2024, 12(10):1448-1454. 

[15] Xu S, Zhang Y, An X, et al. Performance Evaluation of Seven Multi-Label Classification Methods on Real-World Patent and Publication Datasets [J]. Journal of Data and Information Science, 2024, 9(2):81-103.