Prioritizing promising technology domains for technical idea generation through structured problem–domain matching

Nakamura, Nobuhiro; Tomita, Hikari; Kondo, Shinji; Kamiya, Toshio; Terayama, Kei

doi:10.51094/jxiv.6966

##article.authors##

Nakamura, Nobuhiro MDX Research Center for Element Strategy, Institute of Integrated Research, Institute of Science Tokyo
Tomita, Hikari Graduate School of Medical Life Science, Yokohama City University
Kondo, Shinji MDX Research Center for Element Strategy, Institute of Integrated Research, Institute of Science Tokyo
Kamiya, Toshio MDX Research Center for Element Strategy, Institute of Integrated Research, Institute of Science Tokyo
Terayama, Kei 横浜市立大学 https://researchmap.jp/kei_terayama

DOI:

https://doi.org/10.51094/jxiv.6966

キーワード:

Lange Language Models、 Technical idea generation

抄録

Innovative technical solutions often emerge by transferring principles, materials, processes, structures, or functions from technological domains that are distant from the original problem. Recently, the use of large language models (LLMs) for generating candidate solutions using specific knowledge from diverse technological domains has attracted increasing attention. However, identifying promising domains before solution generation remains challenging, and evaluating a large number of candidates generated across a broad set of domains requires substantial effort. Here, we propose a structured problem–domain matching method for prioritizing technology domains for subsequent solution generation. The method represents each problem and candidate domain as a set of functional and technical noun phrases to capture local technical correspondences, and ranks the domains using the upper-tail mean of pairwise embedding similarities. We evaluated the proposed method on two problems: organic light-emitting diode (OLED) light extraction and indium gallium zinc oxide thin-film transistor (IGZO-TFT) source/drain contact formation. Matching-based ranking increased the proportion of domains associated with highly rated solutions in the upper-ranked subsets by more than tenfold relative to the full candidate set in both case studies. These results therefore demonstrate the potential of structured problem–domain matching to render broad-domain solution generation more efficient.

利益相反に関する開示

The authors declare no competing interests.

ダウンロード *前日までの集計結果を表示します

ダウンロード実績データは、公開の翌日以降に作成されます。

引用文献

Christensen, B. T. & Schunn, C. D. The relationship of analogical distance to analogical function and preinventive structure: The case of engineering design. Mem. Cogn. 35, 29–38 (2007).

Linsey, J. S., Wood, K. L. & Markman, A. B. Modality and representation in analogy. AI EDAM 22, 85–100 (2008).

Jiang, S., Hu, J., Wood, K. L. & Luo, J. Data-driven design-by-analogy: State-of-the-art and future directions. J. Mech. Des. 144, 020801 (2022).

Fu, K., Murphy, J., Yang, M., Otto, K., Jensen, D. & Wood, K. L. Design-by-analogy: experimental evaluation of a functional analogy search methodology for concept generation improvement. Res. Eng. Des. 26, 77–95 (2015).

Jing, L., Huang, M., Li, Q., Dou, Y., Feng, D. & Jiang, S. Knowledge graph-assisted design-by-analogy: promoting product innovation through structured analogical knowledge retrieval. Artif. Intell. Eng. Des. Anal. Manuf.: AIEDAM 39, e20, 1–32 (2025).

Luo, J., Sarica, S. & Wood, K. L. Guiding data-driven design ideation by knowledge distance. Knowl.-Based Syst. 218, 106873 (2021).

Fu, K., Chan, J., Cagan, J., Kotovsky, K., Schunn, C. D. & Wood, K. L. The meaning of “near” and “far”: The impact of structuring design databases and the effect of distance of analogy on design output. J. Mech. Des. 135, 021007 (2013).

Sarica, S. & Luo, J. TechNet: Technology semantic network based on patent data. Expert Syst. Appl. 142, 112995 (2020).

Tomita, H., Nakamura, N., Ishida, S., Kamiya, T. & Terayama, K. Extracting effective solutions hidden in large language models via generated comprehensive specialists: case studies in developing electronic devices. Commun. Mater. 6, 207 (2025).

Zhu, Q. & Luo, J. Generative transformers for design concept generation. J. Comput. Inf. Sci. Eng. 23, 041003 (2023).

Zhu, Q., Zhang, X. & Luo, J. Biologically inspired design concept generation using generative pre-trained transformers. J. Mech. Des. 145, 041409 (2023).

Jiang, S., Li, W., Qian, Y., Zhang, Y. & Luo, J. AutoTRIZ: Automating engineering innovation with TRIZ and large language models. Adv. Eng. Inform. 65, 103312 (2025).

Ma, K., Grandi, D., McComb, C. & Goucher-Lambert, K. Do large language models produce diverse design concepts? A comparative study with human-crowdsourced solutions. J. Comput. Inf. Sci. Eng. 25, 024501 (2025).

Si, C., Yang, D. & Hashimoto, T. Can LLMs generate novel research ideas? A large-scale human study with 100+ NLP researchers. In Proceedings of the Thirteenth International Conference on Learning Representations https://proceedings.iclr.cc/paper_files/paper/2025/hash/ea94957d81b1c1caf87ef5319fa6b467-Abstract-Conference.html (2025).

Baek, J., Jauhar, S. K., Cucerzan, S. & Hwang, S. J. ResearchAgent: Iterative research idea generation over scientific literature with large language models. Proc. Conf. Nations Am. Chapter Assoc. Comput. Linguist. Hum. Lang. Technol. 6709–6738 (2025).

Gu, X. & Krenn, M. Generation and human-expert evaluation of interesting research ideas using knowledge graphs and large language models. NeurIPS 2024 Workshop on Machine Learning and the Physical Sciences (2024).

Lewis, P. et al. Retrieval-augmented generation for knowledge-intensive NLP tasks. Proceedings of the 34th International Conference on Neural Information Processing Systems, 9459–9474 (2020).

Siddharth, L. & Luo, J. Retrieval augmented generation using engineering design knowledge. Knowl.-Based Syst. 303, 112410 (2024).

Swanson, K., Wu, W., Bulaong, N. L., Pak, J. E. & Zou, J. The Virtual Lab of AI agents designs new SARS-CoV-2 nanobodies. Nature 646, 716–723 (2025).

Ghafarollahi, A. & Buehler, M. J. SciAgents: Automating scientific discovery through bioinspired multi-agent intelligent graph reasoning. Adv. Mater. 37, 2413523 (2025).

Boiko, D. A., MacKnight, R., Kline, B. & Gomes, G. Autonomous chemical research with large language models. Nature 624, 570–578 (2023).

Bran, A. M., Cox, S., Schilter, O., Baldassari, C., White, A. D. & Schwaller, P. Augmenting large language models with chemistry tools. Nat. Mach. Intell. 6, 525–535 (2024).

Kang, Y. & Kim, J. ChatMOF: an artificial intelligence system for predicting and generating metal-organic frameworks using large language models. Nat. Commun. 15, 4705 (2024).

Chaudhari, A., Ock, J. & Barati Farimani, A. Modular large language model agents for multi-task computational materials science. Commun. Mater. 7, 131 (2026).

Nakamura, N. et al. 58.1: Invited Paper: Advanced glass substrate for the enhancement of OLED lighting out-coupling efficiency. Dig. Tech. Pap. - SID Int. Symp. 44, 803–806 (2013).

Shi, Y. et al. Approach to low contact resistance formation on buried interface in oxide thin-film transistors: utilization of palladium-mediated hydrogen pathway. ACS Nano 18, 9736–9745 (2024).

World Intellectual Property Organization. Guide to the International Patent Classification 2025. WIPO/GUIDE/IPC/2025/EN (2025).

OpenAI. GPT-5.6 system card. https://deploymentsafety.openai.com/gpt-5-6 (2026).

OpenAI. Update to GPT-5 system card: GPT-5.2. https://openai.com/index/gpt-5-system-card-update-gpt-5-2/ (2025).

Lee, J. et al. Gemini embedding: Generalizable embeddings from Gemini. arXiv:2503.07891 (2025).

OpenAI. GPT-4o system card. https://openai.com/index/gpt-4o-system-card/ (2024).