A Multi-Subject Differentiated Patent Recommendation Algorithm Integrating Knowledge Graphs

DOI: https://doi.org/10.62517/jike.202604114

Author(s)

Hongxia Zhang¹, Jian Ma2*

Affiliation(s)

1Yunnan Academy of Scientific & Technical Information, China 2Department of Information System, City university of Hong Kong, China *Corresponding Author

Abstract

Patent recommendation systems are crucial decision-support tools for innovation management, yet existing algorithms often neglect the heterogeneous needs of stakeholders (enterprises focus on market value, research institutions on technical novelty, and patent attorneys on legal risks). Conventional models adopt homogeneous ranking strategies that fail to align with differentiated decision goals. To address this gap, this study proposes a multi-subject differentiated patent recommendation algorithm integrating domain-specific knowledge graphs and demand-weighted graph neural networks (DW-GNN). First, a three-dimensional stakeholder demand framework (Technical, Value, Risk) is defined based on systematic patent valuation literature, and subject-specific weights are calculated via a hybrid AHP-entropy method. A demand-weighted graph model learns representation vectors incorporating stakeholder priorities, and a multi-objective scoring function generates subject-adapted rankings. Experiments show that the algorithm improves recommendation performance across stakeholder groups using standard evaluation metrics. This research contributes a stakeholder-oriented patent analytics system, advancing personalized knowledge graph reasoning theory and supporting innovation management practice.

Keywords

Patent Recommendation; Knowledge Graph; Graph Neural Network; Multi-Subject Modeling

References

[1]Aleman-Meza, B., Halaschek-Wiener, C., Arpinar, I. B., Ramakrishnan, C. R., & Sheth, A. (2005). Ranking complex relationships on the semantic web. IEEE Internet Computing, 9(3), 37-44. https://doi.org/10.1109/mic.2005.63 [2]Chen, H., & Deng, W. (2023). Interpretable patent recommendation with knowledge graph and deep learning. Scientific Reports, 13(1). https://doi.org/10.1038/s41598-023-28766-y [3]Dehdasht, G., Ferwati, M. S., Zin, R. M., & Abidin, N. Z. (2020). A hybrid approach using entropy and TOPSIS to select key drivers for a successful and sustainable lean construction implementation. PLOS ONE, 15(2), e0228746. https://doi.org/10.1371/journal.pone.0228746 [4]Du, W., Yan, Q., Zhang, W., & Ma, J. (2021). Leveraging online behaviors for interpretable knowledge-aware patent recommendation. Internet Research, 32(2), 568-587. https://doi.org/10.1108/intr-08-2020-0473 [5]Érdi, P., Makovi, K., Somogyvári, Z., Strandburg, K. J., Tobochnik, J., Volf, P., & Zalányi, L. (2012). Prediction of emerging technologies based on analysis of the US patent citation network. Scientometrics, 95(1), 225-242. https://doi.org/10.1007/s11192-012-0796-4 [6]Freunek, M., & Bodmer, A. (2021). BERT based freedom to operate patent analysis. arXiv. https://doi.org/10.48550/arxiv.2105.00817 [7]Jeon, J., & Suh, Y. (2019). Multiple patent network analysis for identifying safety technology convergence. Data Technologies and Applications, 53(3), 269-285. https://doi.org/10.1108/dta-09-2018-0077 [8]Lai, Y., Che, H., & Wang, S. (2008). Managing patent legal value via fuzzy neural network incorporated with factor analysis based on patent infringement lawsuits. 2008 4th International Conference on Wireless Communications, Networking and Mobile Computing. https://doi.org/10.1109/wicom.2008.2245 [9]Lee, J. (2020). Controlling patent text generation by structural metadata. Proceedings of the 29th ACM International Conference on Information & Knowledge Management, 3241-3244. https://doi.org/10.1145/3340531.3418503 [10]Lee, J., & Hsiang, J. (2020). PatentTransformer-1.5: Measuring patent claim generation by span relevancy. Lecture Notes in Computer Science, 20-33. https://doi.org/10.1007/978-3-030-58790-1_2 [11]Lin, L. (2011). Licensing strategies in the presence of patent thickets. Journal of Product Innovation Management, 28(5), 698-725. https://doi.org/10.1111/j.1540-5885.2011.00835.x [12]Liu, Q., Wu, H., Ye, Y., Zhao, H., Liu, C., & Du, D. (2018). Patent litigation prediction: A convolutional tensor factorization approach. Proceedings of the Twenty-Seventh International Joint Conference on Artificial Intelligence, 5052-5059. https://doi.org/10.24963/ijcai.2018/701 [13]Magdy, W., & Jones, G. J. F. (2011). An efficient method for using machine translation technologies in cross-language patent search. Proceedings of the 20th ACM International Conference on Information and Knowledge Management, 1925-1928. https://doi.org/10.1145/2063576.2063856 [14]Mahdabi, P., & Crestani, F. (2014). Patent query formulation by synthesizing multiple sources of relevance evidence. ACM Transactions on Information Systems, 32(4), 1-30. https://doi.org/10.1145/2651363 Mauleón, A., Vannetelbosch, V., & Vergari, C. (2013). Bargaining and delay in patent licensing. International Journal of Economic Theory, 9(4), 279-302. https://doi.org/10.1111/j.1742-7363.2013.12020.x [15]Mediani, C. (2022). Interactive hybrid recommendation of pedagogical resources. Ingénierie Des Systèmes D Information, 27(5), 695-704. https://doi.org/10.18280/isi.270502 [16]Nguyen, T., Le-Cong, T., Nguyen, T., Le, X. D., & Thắng, H. Q. (2022). Toward the analysis of graph neural networks. arXiv. https://doi.org/10.48550/arxiv.2201.00115 [17]Nyimbili, P. H., & Erden, T. (2020). A hybrid approach integrating entropy-AHP and GIS for suitability assessment of urban emergency facilities. ISPRS International Journal of Geo-Information, 9(7), 419. https://doi.org/10.3390/ijgi9070419 [18]Österman, J. (2018). Non-discriminatory royalties for standard essential patents: Setting the boundaries. Nordic Journal of European Law, 1(1), 24-62. https://doi.org/10.36969/njel.v1i1.18680 Parulian, D., Fazrie, M., & Alima, G. (2023). Decision support system for selecting tofu dregs suppliers using the analytical hierarchy process (AHP) method. International Journal Software Engineering and Computer Science (IJSECS), 3(3), 487-494. https://doi.org/10.35870/ijsecs.v3i3.1935 [19]Priya, R. V., & Vadivel, A. (2012). Capturing semantics of web page using weighted TAG-tree for information retrieval. International Journal of Asian Business and Information Management, 3(4), 7-24. https://doi.org/10.4018/jabim.2012100102 [20]Sangka, K. B., & Muchsini, B. (2018). Accommodating analytic hierarchy process (AHP) for elective courses selection. IJIE (Indonesian Journal of Informatics Education), 2(2). https://doi.org/10.20961/ijie.v2i2.24436 [21]Scrivano, A. (2025). Graph neural networks: Techniques and applications for data structured as graphs. arXiv. https://doi.org/10.31224/5145 [22]Siddharth, L. (2025). Modularizing patent knowledge for enhanced technological impact. arXiv. https://doi.org/10.31219/osf.io/fd36m_v3 [23]Song, B., & Luo, J. (2017). Mining patent precedents for data-driven design: The case of spherical rolling robots. Journal of Mechanical Design, 139(11). https://doi.org/10.1115/1.4037613 Starešinič, M., & Boh, B. (2009). Patent informatics: The issue of relevance in full‐text patent document searches. Online Information Review, 33(1), 157-172. [24]Stoffels, M., Klauck, F. J. R., Hamadi, T., Glorius, F., & Leker, J. (2020). Technology trends of catalysts in hydrogenation reactions: A patent landscape analysis. Advanced Synthesis & Catalysis, 362(6), 1258-1274. https://doi.org/10.1002/adsc.201901292 [25]Verma, M., & Varma, V. (2011). Patent search using IPC classification vectors. Proceedings of the 4th Workshop on Patent Information Retrieval, 9-12. https://doi.org/10.1145/2064975.2064980 [26]Wang, F., Lin, L. F., & Zhou, Y. (2013). An ontology-based automatic semantic annotation approach for patent document retrieval in product innovation design. Applied Mechanics and Materials, 446-447, 1581-1590. https://doi.org/10.4028/www.scientific.net/amm.446-447.1581 [27]Wu, S., Sun, F., Zhang, W., Xie, X., & Cui, B. (2020). Graph neural networks in recommender systems: A survey. arXiv. https://doi.org/10.48550/arxiv.2011.02260 [28]Yang, G., Li, G., Li, C., Zhao, Y., Zhang, J., Liu, T., & Huang, M. (2015). Using the comprehensive patent citation network (CPC) to evaluate patent value. Scientometrics, 105(3), 1319-1346. https://doi.org/10.1007/s11192-015-1763-7 [29]Zarrinkalam, F., & Kahani, M. (2013). SemCiR: A citation recommendation system based on a novel semantic distance measure. Program, 47(1), 92-112. [30]Zhao, L., Luan, N., Cheng, W. K., Feng, S., Wang, H., Yang, Y., & Zhu, G. (2025). Intent-aware knowledge graph-based model for electrical power material recommendation. PeerJ Computer Science, 11, e3023. https://doi.org/10.7717/peerj-cs.3023