Logo image
以主成份分析與深度神經網路為基之專利品質評估方法論:以物聯網應用於製造業為案例
Thesis

以主成份分析與深度神經網路為基之專利品質評估方法論:以物聯網應用於製造業為案例

孫俊弘
Masters, 國立清華大學, 工業工程與工程管理學系所
2016

Abstract

專利指標 專利品質 專利商品化 主成份分析法 深度神經網路 patent indicators patent quality patent commercialization principal component analysis deep neural networks
In the era of knowledge economy and industry 4.0, global production costs for products have become increasingly competitive, how companies obtain benefits of patent commercialization by developing key technologies to make applications for related patents becomes very important. However, a large quantity of patent information requires much time for enterprises to identify patent values. Traditional evaluation methods of patent quality rely on experts to determine patent values, which requires long time spent and ends with subjective assessment results. Therefore, the research proposes a quantitative evaluation method of patent quality by constructing a Python-based system for patent quality classification to research on patent evaluations. Evaluation of patent values by taking patent indicators into consideration helps explore what kind of patents are fundamental and essential for enterprises. Patent evaluations help managers and R&D personnel understand patent values and identify related business opportunities, which further determines the directions of patented technology development. The process for the research method is divided into three parts. The first part is patent retrieval and patent indicators collections. Second part introduces Principal Component Analysis (PCA) to make key patent indicators selection. Third part ends with the Deep Neural Networks (DNN) analysis result. The research takes Internet of Things (IoT) applied in manufacturing industries as the example case, and find out high-value patents by using DNN to classify patent quality. Moreover, the research puts many efforts into both worldwide high-value patents and those of Taiwan to explore their portfolios, which provides useful suggestions for Taiwan assignees in the R&D strategies of IoT technologies.

Metrics

1 Record Views

Details

Logo image