An Analysis of the Patent Search Behaviour of Human Experts and Its Impact on Patent Retrieval Performance

Abstract

Background: Patent databases are essential repositories of technical information that support innovation and technological advancements across industries. Despite their value, retrieving relevant patents from these databases remains time-consuming and often inaccurate. This challenge is exacerbated by the fact that human experts essentially perform patent retrieval, whose search approaches and skill levels vary significantly. These variations lead to inconsistent outcomes, making the retrieval process unreliable and inefficient. Understanding the factors that influence patent retrieval performance is critical for improving the utility of these databases and facilitating the discovery of enabling technologies. Objectives: This study aims to investigate the impact of human search behaviour on patent retrieval performance. By analysing specific aspects of expert behaviour, the research seeks to identify factors contributing to variability in retrieval outcomes and propose strategies for improvement. The study focuses on keyword diversity, query complexity, and search speed metrics to measure search behaviour while evaluating reliability, efficiency, effectiveness, and judgement error. Ultimately, the objective is to assess whether current human-centred approaches are sufficient and explore potential solutions for enhancing retrieval. Main Ideas: The study observes seven patent retrieval experts searching for, extracting, and evaluating metallurgical patents containing enabling technologies for developing and producing an innovative kitchen skillet. Through systematic analysis, the researcher identifies patterns in search behaviour and correlates them with retrieval performance. The findings reveal a strong relationship between search strategies and outcomes, with significant variations in performance across the experts. Metrics such as keyword diversity and query complexity emerge as key determinants of retrieval success. However, the results highlight that the overall performance of human experts in patent retrieval is generally low, demonstrating inefficiencies and inconsistencies in their approaches. Conclusion: The results of this study underscore the limitations of current human-centred patent retrieval methods and the need for more effective approaches. The significant variability in expert performance suggests that relying solely on human expertise is insufficient for achieving consistent and accurate results. Applying artificial intelligence (AI) methods to patent retrieval has the potential to address these challenges by automating key aspects of the search process and improving both efficiency and accuracy. By integrating AI tools, organisations can significantly enhance patent retrieval performance and unlock the full value of patent databases.

Keywords: Patent Search, Patent Retrieval, Search Behaviour, Retrieval Performance

View PDF