Research On Patent Lifespan Prediction And Influencing Factors

Scientific and technological innovation plays a critical role in economic development.With the rapid growth of economies worldwide,competition in scientific and technological innovation has intensified.In this fiercely competitive patent environment,improving the quality and value of patents to protect innovation achievements and enhance the scientific and technological competitiveness of enterprises and countries has become an important research topic in practical and theoretical fields.Patent lifespan is a key indicator for evaluating the merits of the patent system.It reflects the system’s performance,evaluates the patent management and application abilities of innovation entities,and assesses technological innovation capabilities.Studying the influencing factors and prediction models of patent lifespan is not only of the great theoretical value for improving the patent system,particularly the patent maintenance system,but also of significant practical significance for enhancing the patent application and management abilities of innovation entities and improving the core competitiveness of countries or regions.This paper focuses on the analysis and prediction of factors affecting the patent lifespan,using patent literature as the basis and employing methods such as bibliometrics,text mining,machine learning,and survival analysis.The framework comprises a theoretical framework with a methodological basis to: analysis of factors affecting the patent lifespan;construction of a patent lifespan prediction model;identification of potentially high-value patents from the perspective of patent lifespan.A feasible empirical analysis framework based on the algorithmic attribution perspective for the factors affecting patent lifespan,a more accurate patent lifespan prediction model,and a reliable method for identifying potentially high-value patents from the perspective of patent lifespan have been developed.The main content of this thesis is summarized as follows.Firstly,the paper proposes an empirical research framework for analyzing the impact factors of patent lifespan using algorithmic attribution.The study defines the analysis of patent lifespan impact factors as a survival analysis problem due to the limitations of existing methods.The framework comprises six processes,including constructing a patent lifespan impact factor index system,developing and evaluating a benchmark prediction model,constructing an explainer using Shapley’s theory,conducting exploratory analysis of impact factors based on a full sample,and testing the robustness of the attribution framework.The paper summarizes and establishes an index system of factors that influence patent lifespan from a biological perspective.Using data from invention patents granted in China from 2001-2017,the study employs an algorithmic attribution approach to examine the influencing factors of patent lifespan from multiple dimensions.The findings indicate that innate factors have the largest influence on patent lifespan,followed by behavioral and environmental factors.The impact of different factors on patent lifespan is moderated by the patent literature life cycle and the technology field.The paper summarizes and establishes an index system of factors that influence patent lifespan from a biological perspective.Using data from invention patents granted in China from 2001-2017,the study employs an algorithmic attribution approach to examine the influencing factors of patent lifespan from multiple dimensions.The findings indicate that innate factors have the largest influence on patent lifespan,followed by behavioral and environmental factors.The impact of different factors on patent lifespan is moderated by the patent literature life cycle and the technology field.Secondly,a patent lifespan prediction model based on multi-feature fusion is constructed.In view of the problem that patent metrics can only show the external features of patent documents but not their internal features,the thesis proposes a patent lifespan prediction model based on multi-feature fusion,which fuses the external attribute features of patents with the semantic features of the text of patent documents.The proposed model consists of four steps: data collection,data pre-processing,feature extraction and fusion,model construction and evaluation.Furthermore,the paper conducts empirical research using invention patents in the field of OLED technology authorized by the China National Intellectual Property Administration as samples.The cross-validation and Monte Carlo simulation results show that the patent lifespan prediction model based on multi-feature fusion has good performance in CI,IBS and INBLL values,indicating that the patent lifespan prediction model constructed in this paper can predict patent survival risk and patent lifespan more accurately.Thirdly,the paper proposes a method for identifying potential high-value patents from the perspective of patent lifespan.To address the question of how to apply the patent lifespan prediction model,the study analyzes the feasibility of identifying potential highvalue patents from the perspective of patent life,and presents a method that includes data collection,data processing,patent lifespan prediction model construction,and potential high-value patent identification.To verify the feasibility of the proposed method,the paper conducts an empirical study using patents in the field of Augmented Reality technology as samples.The method is forward-looking and can identify potential high-value patents during the patent application confirmation stage.The proposed method offers a valuable contribution to the field and is worthy of further promotion and adoption.This paper provides a comprehensive review and summary of relevant domestic and foreign research on patent lifespan.Drawing an analogy between patent lifespan and natural biological life span,the study defines the analysis and prediction of patent lifespan as a survival analysis problem.The paper presents an analysis framework of patent lifespan influencing factors based on algorithmic attribution and a patent lifespan prediction model.The feasibility,reliability,and accuracy of the patent lifespan prediction model are investigated to identify potential high-value patents.The research results offer solutions for investigating the influencing factors of patent lifespan and establishing more accurate patent lifespan prediction models and potential high-value patent identification models.Furthermore,the study provides new research perspectives,methods,and models for patent lifespan-related research,patent intelligence analysis,and high-value patent screening.Overall,this paper makes a valuable contribution to the field of patent research and offers practical insights for industry professionals and researchers alike.