Research On Chinese-english Bilingual Patent Information Retrieval And Topic Clustering

Patent literature is a huge source of information,which is an important form of expression of the science and technology research achievements,can reflect the achievements and the latest development of scientific and technological.With the rapid development of China's market economy and the deepening of economic globalization,the conflict of patent and patent barriers deeply affects the broad masses of Chinese enterprises and research institutes,the patent infringement and patent protection has become the focus,as well as the research and the use of foreign patent.This paper focus on the Chinese-English bilingual patent retrieval,help patent analysts collecting the patent literature accurately wherever domestic or overseas by means of bilingual retrieving,and made the study on the bilingual patent topic clustering based on the evolution of topic,to analyse the distribution and evolution of the bilingual topic,finally,according to the actual demand,developing the system prototype implemented bilingual retrieval and topic clustering of patent,and give a example with bilingual patent in the field of"3d printing".This paper proposed a bilingual patent retrieve scheme through building base bilingual dictionary,professional bilingual dictionary and bilingual parallel patent corpus with mutual complementation relationship,to support accurate translation by buliding bilingual space of patent to resolve ambiguity resolution problem.Using the vector space model to represent the patent literature index on title and abstract,finally building the patent retrieval scheme and conducting a small-scale retrieval experiments in the "3d printing" field to verify the feasibility of the bilingual retrieval scheme and the availability of the ambiguation method.On the base of former study,made topic clustering analysis with the HDP topic model mainly on the patent title and abstract,which can represented patent text with the probability distribution of a set of topics to mining potential technical topic,furthermore,divided patent collection by apply time to make topic cluster respectively,by which can analyse patent topic distribution and evolution,finally made an instance analysis with "3d printing" bilingual patents.Finally,from the perspective of practical application,developed Chineses-English bilin-gual patent retrieval and topic clustering analysis system based on the J2EE platform.The system implement the functions such as Chinese-English bilingual patent information retrieval,the Chinese-English bilingual patent topic clustering and dictionary management,and we display the topic clustering results with the form of visual charts.