專利題目擬訂推薦電腦輔助方法與系統

Patent Title Develop Recommending Computer-Aided Method And System

由於資訊科技的快速發展下，知識透過資訊在全球化下快速的擴散，現今全球各地的企業公司正朝向知識與研發密集化的趨勢逐漸發展，而這些的知識與研發密集化的企業公司之發展攸關一個國家的經濟、科技政策，並非一家或少數幾家企業所能獨力推展，都是需要透過國家以整體力量在政策上來給予協助。而透過上述知識與研發所衍生的經濟價值，將成為技術的創新與研發的主要核心要素，在專利市場上的佈局策略與智慧財產所衍生出來的價值應用將成為最關鍵性的角色。本系統最主要的目標為透過大量的專利資料庫、詞典本體論知識庫的建置，並藉由語意網與運用R語言－自然語言處理：關鍵詞提取（TF-IDF）來作詞頻分析，以達到能根據使用者對於專利上的需求來產生相對應的專利文件名稱。並運用權重技術針對專利內容進行文字探勘後，將其進行分析詞彙的重要性，再運用這些詞彙建置詞典本體論，運用在本系統的關鍵字擴展，當使用者透過輸入的關鍵字後，本系統會將輸入的關鍵字組合處理透過權重的排序，給於使用者推廌，以達到能根據使用者對於專利上的需求來產生相對應的專利文件名稱。且本系統透過原專利的題目來做驗證，驗證後超過6成的相似度，所以本系統是能夠透過分析來幫助使用者理解知識的內容。 With the rapid development of information technology, knowledge is spreading rapidly through the globalization of information. Today's companies around the world are moving towards the trend of knowledge and R&D (research and development) intensity. This knowledge and the development of intensive corporate companies may affect a country's economic, scientific and technological policies. It is not the one or some companies that can carry out by their own efforts. It is necessary to provide assistance through the government's overall strength in policy. However, through the above-mentioned knowledge and the economic value generated by R&D., it will become the main core element of technology innovation and R&D. The layout strategy in the patent market and the application of value generated by intellectual property will become the most critical role. The main goal of this system is to build a large scale of patent database and dictionary ontology knowledge base. And through the semantic network and the use of R language-natural language's keyword extraction (TF-IDF) for word frequency analysis, a corresponding patent title can be generated according to the users' demand for their patents. After using the weighting technology to conduct a text search on the patent content, we analyze the importance of the vocabulary and then use these words to build the dictionary ontology. In the use of the system's keyword extension and users' input of keywords, the system will process the input keyword combination through the weight sorting and then give the user a list of recommended title. This can meet the system users' demand to generate the corresponding patent title. The system was verified by the title of the previous published patent. It has more than 60% similarity after verification. Therefore, this system is a useful tool that can help users understand knowledge and generate suitable patent title through computer-aided analysis.