The Study Of Eye-Movement And ERP Feature In Users' Patent Retrieval And Its Application Of Retrieval Reordering

With the development of digital era,digital information increases exponentially,leading to an increasing number of users' search results for the same problem.The amount of redundant search results obtained by users reduces the accuracy of retrieval.So,how to improve the rationality of retrieval results ranking becomes the key issue in this field.This study,taking user patent retrieval as an example,excavated users' interest and optimized the retrieval results according to user's interest.By improving the satisfaction of users' search results with questionnaires,semi structured interviews,eye movements and ERP experiments.This study mainly includes the following aspects.(1)Using the method of literature research,analysis on the information behavior research and optimization of retrieval results around the world.Combing the existing research on information behavior around the world,find frontiers of this field.Experimental methods have good research foundation and theoretical basis in information behavior study.Reordering retrieval results is mainly based on the mining of user data.This thesis compares the differences of literature,provides inspiration for the optimization of patent retrieval results.(2)The study of the calculation of the degree of interest by using user's feature of eye-movement and click behavior in patent retrieval.Combined with the experiment and questionnaire interview method,the process of reading and retrieving results of 24 subjects was recorded and analyzed by Tobii T60 XL eye tracking system.Through the analysis of interest degree and user behavior data,the general method of interest expression of search results has produced,which provides the basis for retrieval ranking optimization.(3)Research on ERP(event-related potential)behavior feature of user patent retrieval and its application in interest calculation.The relationship between ERP behavior feature and users' interest in retrieval results is explored and verified.Through the hypothesis-verification experimental paradigm,the relationship between the users' interest in the retrieval results and the ERP components is proposed.Through the EGI ERP recording equipment,24 subjects are tested to verify the feasibility of using the ERP signal to express the interest of the user.Experimental results are discussed to calculate the interest degree of users.(4)Empirical research on ranking optimization of patent retrieval results.By analyzing the feature of patent documents,a ranking optimization method based on the feature of patent documents is constructed.The international patent classification(IPC)of the patent literature joint with the existing information of users,and the dynamic data of the user behavior collected by the experiment are combined to optimize the results of the patent retrieval.Through the empirical analysis of 24 testee,23.7% of the retrieval results are presented to the user at least one page in advance,which improves the efficiency of user retrieval.The testee data obtained through the questionnaire also confirm the optimization effect.