Modeling the Dynamic Process of Inventions for Reducing Knowledge Search Costs

Haiying Ren,Yuanyuan Song,Rui Peng

2024-05-10

Abstract:A knowledge search is a key process for inventions. However, there is inadequate quantitative modeling of dynamic knowledge search processes and associated search costs. In this study, agent-based and complex network methodologies were proposed to quantitatively describe the dynamic process of knowledge search for actual inventions. Prior knowledge networks (PKNs), the search space of historical patents, were constructed, representative search rules were formulated for R&D agents, and measures for knowledge search cost were designed to serve as search objectives. Simulation results in the field of photolithographic technology show that search costs differ significantly with different search rules. Familiarity and Degree rules significantly outperform BFS, DFS and Recency rules in terms of knowledge search costs, and are less affected by the size and density of PKNs. Interestingly, there is no significant correlation between the mean and variance of search costs and patent value, indicating that high-value patents are not particularly difficult to obtain. The implications for innovation theories and R&D practices are drawn from the models and results.

Social and Information Networks

What problem does this paper attempt to address?

### Problems the Paper Attempts to Solve

This paper aims to address the issue of dynamic modeling in the knowledge search process and the quantification of its associated costs. Specifically, the authors propose an agent-based and complex network approach to quantitatively describe the dynamic process of knowledge search in actual invention. The paper focuses on the following aspects:

1. **Cost Modeling of Knowledge Search**:
- Knowledge search is a critical process in invention, but there is currently a lack of quantitative modeling of the dynamic knowledge search process and its costs.
- The authors construct an agent-based dynamic model to describe the knowledge search in the invention process and design metrics to measure the cost of knowledge search.

2. **Effects of Different Search Rules**:
- By simulating the application of different search rules (such as familiarity, degree, recency, etc.) in the field of lithography technology, the study examines the impact of these rules on the cost of knowledge search.
- It is found that the familiarity and degree rules significantly outperform breadth-first search (BFS), depth-first search (DFS), and recency rules in terms of knowledge search cost and are less affected by the size and density of the knowledge network.

3. **Relationship Between Knowledge Search Cost and Patent Value**:
- The paper explores the relationship between knowledge search cost and patent value, finding that obtaining high-value patents is not particularly difficult, i.e., the search cost for high-value patents is not significantly higher than that for low-value patents.

### Main Contributions

- **Methodological Innovation**: Proposes an agent-based and complex network approach to quantitatively describe the dynamics of knowledge search in the invention process.
- **Empirical Analysis**: Validates the effects of different search rules using actual patent data in the field of lithography technology and provides specific cost comparisons.
- **Theoretical Significance**: Offers a new perspective for innovation theory and R&D practice, emphasizing the importance of optimizing knowledge search strategies to improve innovation efficiency and reduce costs.

### Conclusion

By constructing and simulating a knowledge search model, the paper reveals the impact of different search rules on the cost of knowledge search and explores the relationship between knowledge search cost and patent value. The research results indicate that choosing appropriate search rules can significantly reduce the cost of knowledge search, thereby improving innovation efficiency. These findings have important implications for innovation management and R&D practice.