Abstract: A new method of text mining was presented to make up for the disadvantages of big data knowledge acquisition.Firstly,we constructed triple ontology model about academic inventive features “problem,methods,results”.Se-condly,pattern recognition techniques were used for statistical analysis,feature extraction,machine learning and pattern determination analysis.Finally,depth mining of triples of the creative core academic knowledge was realized.The Expe-rimental results show that the new method can effectively reduce the retrieval noise of academic literature,which is convenient for users to quickly locate the research problem.The methods and results can determine the reading value of papers and provide a basis for depth knowledge mining of large data and related discovery.The method has not been reported in the literature,and it is a kind of exploratory research and experimentation.

Key words: Pattern recognition,Text mining,Semantic triples,Direct knowledge acquisition