Research On Topic Mining And Application Of Auto Patent Text Based On Topic Model

Auto patent text,with its professional and high technical value,provides an important way for auto manufactures or related researchers to grasp the industry’s technological development,seek technological innovation,etc.Existing research and analysis of auto patent texts are always based on traditional patent measurement analysis,text mining models or methods,etc.They either make rough statistical analysis based on their structured parts,or can only dig out the shallow information of their unstructured text content.These models or methods are rarely able to dig deeper into the semantic information of auto patent texts,which cause it can not make some detailed analysis or applications from semantic perspective.With the application and development of Natural Language Processing technology and related machine learning models and methods,the topic model represented by LDA model has great advantages in the content analysis of unstructured text,and has been widely applied in many fields of text mining tasks.The topic model reveals the text’s semantic information by extracting the topic embodied in the text,achieves better representation of the text content.Therefore,based on the status of auto patent text research and analysis,this thesis carries out the research of its topic mining and application based on topic model.The main work completed is as follows:(1)Constructed a framework for auto patent text topic mining and analysis.On the basis of analyzing the composition,characteristics,IPC classification and common analysis indexes of auto patent text,this thesis constructed a topic-model-based mining and analysis framework for its unstructured text content mining and analysis.(2)Researched on topic mining method based on topic model for auto patent text.On the basis of analyzing the advantages and disadvantages of several common patent text topic mining methods,a topic mining method based on IPC classification number and LDA model is proposed,and an improved hot topic recognition method is given,to get fine-grained topic mining and hot topic recognition for auto patent text.(3)Designed and implemented a topic mining and analysis system based on topic model for auto patent text.The objectives and requirements of the system are analyzed and the system architecture is designed.The main function modules,such as text preprocessing module,text topic mining module,hot topic recognition module,are designed and implemented in detail,and the actual results of their operation are tested.