Research On Reminding Algorithms For Creative Design Support Engine

With economic globalization, competition among enterprises has become increasingly intense. Innovation is the key to improving the competitiveness of enterprises, which has become a consensus. As a senior human thinking activity, innovation is difficult. On one hand, innovation requires that people spend years to master a wealth of domain knowledge. On the other hand, when facing with new problems, most people unconsciously tend to think in an existing, familiar and narrow framework, such thinking inertia and mindset becomes the major obstacle to innovation. In order to accelerate the innovation process and help people out of the bondage of habitual thinking, computer aided thinking has aroused widespread interest in the study.The paper first briefly explains the importance and difficulties of innovation, analyses the necessity and research status of computer aided thinking, and then introduces the computational creativity theory. Based on this theory, this paper designs a B/S structured prototype system of computer aided thinking—CDSE(Creative Design Support Engine), and defines the main framework and related workflows. CDSE consists of four main modules: problem description, reminding process, creating process and evaluating process. This paper focuses on discussing the implementation of reminding process. According to text similarity and normal distribution principle, this system searches patents logically related to specific problem to form the reminding space. Finally, experimental results prove that reminding algorithms proposed in this paper are effective and can provide some good reminding materials for creating process of CDSE.The implementation of reminding process is an exploratory work of simulating human associative thinking, it is beneficial to get rid of habitual thinking and provide knowledge for creating process in CDSE. During the implementation of reminding process, the relativity computing between existing patents and specific problem is the key part. Therefore, after analyzing the structural features of patent text, this paper discusses the method of measuring logical relativity among patents. CDSE uses semantic similarity as metric, defines logical relativity based on patent title and patent abstract similarity and reduces the search space through patent IPC.