Research On Technology Novelty Measurement Based On Knowledge Combination Perspective

Technological novelty is regarded as an important driving force for breakthrough technology.However,the concept of technological novelty is abstract and complex.How to measure technological novelty has become a general concern of scholars in recent years.This study proposes a comprehensive measurement index to identify novel technologies from the perspective of knowledge combination.The comprehensive index from knowledge(respectively by patent IPC classification code and patent topic content representation)in the past the co-occurrence frequency,the probability of future link and semantic level similarity measure knowledge from three aspects of the combination of novelty value.Using the method of novelty measurement published in Science journals,the 10th percentile/median of the novelty value distribution of the patent knowledge element combination is used to represent the novelty/conventionality of the patent respectively,and the types of patents were divided into the following types:high conventionality/low novelty,high conventionality/high novelty,low conventionality/high novelty,and low conventionality/low novelty.Using invention patents in the field of artificial intelligence as empirical verification data,research shows that the method of measuring technological novelty proposed by this research can capture more novelty knowledge combinations.Combining the three indicators of past co-occurrences,future link probability and semantic similarity of knowledge elements has certain advantages compared with using a single indicator.In addition,this study proposes that the accuracy of the comprehensive measurement index that combines the IPC combination and the subject word combination is 0.943,indicating the reliability of the comprehensive method to measure the novelty of the technology.Regardless of whether the technical novelty is measured from the IPC code or the subject content,or from the combination of these two aspects,research shows that teamwork has become the mainstream mode,which has a driving effect on the generation of high-novelty patents.High-novelty/high-conventional patents have a high average number of citations and value.Patents with high-novelty features are most likely to develop into high-impact patents.However,people will delay the identification of novel knowledge combinations.Highly cited patents are not necessarily high-novelty patents,indicating that high-novelty patents have the characteristics of high impact and high risk.