The generation of problem-focussed patent clusters: a comparative analysis of crowd intelligence with algorithmic and expert approaches

Abstract

This paper presents a new crowdsourcing approach to the construction of patent clusters, and systematically benchmarks it against previous expert and algorithmic approaches. Patent databases should be rich sources of inspiration which could lead engineering designers to novel solutions for creative problems. However, the sheer volume and complexity of patent information means that this potential is rarely realised. Rather than the keyword driven searches common in commercial systems, designers need tools that help them to understand patents in the context of the problem they are considering. This paper presents an approach to address this problem by using crowd intelligence for effective generation of patent clusters at lower cost and with greater rationale. A systematic study was carried out to compare the crowd's efficiency with both expert and algorithmic patent clusters, with the results indicating that the crowd was able to create 80% more patent pairs with appropriate rationale.

Patent Data for Engineering Design: A Review

“…3, most patent-for-design literature aimed at developing new design methodologies. The earliest work relied on human expertise and manual efforts to extract design rules or heuristics from patents or use patents as design cases to illustrate design methods [11,15,16,17,24,25,26,27,20,28,19,29,30]. More recent studies used NLP to identify innovative solutions from patent database to design problems to facilitate the use of TRIZ [31,32,33,34].…”

“…Koza [51] developed a genetic programming algorithm to solve design problems automatically and utilized patent database to examine the novelty of newly generated solutions. Wodehouse et al [29] presented a clustering method to analyse design opportunities using crowd intelligence. Song, Luo, et al [52] proposed a data-driven product platform design method based on the core-periphery structure detection within functional word co-occurrence networks that are created from patent texts.…”

“…Human subject experiments were primarily employed to develop theory or understanding. In these human studies, full patent documents were often directly provided to designers as design aids such as prior art solutions [11,24,26,25,27,29,30], or design stimuli [15,16,17,20,28,19] for engineering design. Besides, a group of qualitative studies adopted knowledge-based rules or strategies to boost computer-aided engineering design and developed rule-based expert systems [43,31,54,68,42,47,66,33,48,57,45,62,28,13,46,41,12,35].…”

Review of variable leading-edge patents for aircraft wings and engine inlets and their relevance for variable pitot inlets in future supersonic transport

“…Numerous models for the structured classification of patents exist [117]. In the following, the primary differentiation is based on the spatial dimension of the application and the subsequent subdivisions are made according to the associated component and the type of geometry variation, compare Fig.…”

Automated Functional Analysis of Patents for Producing Design Insight

“…Studying patents should help designers to both learn whether their design/invention infringes existing unexpired patents (Ulrich and Eppinger, 2016), and also significantly influence their innovation (Koh, 2020). Studying design concepts and their physical realisations (Li et al, 2012) makes patents a promising source of design inspiration to enhance ideation (Wodehouse et al, 2017(Wodehouse et al, , 2018. However, designers do not often engage with patents, especially in the early stages of design, as analysis of patents requires considerable effort and expertise to extract the technical information buried within the legal terminologies used (Kim et al, 2008;Tseng et al, 2007).…”