The Construction Of Microscopic Concept And The Mining Of Research Ideas In Scientific Papers

Innovation is the inexhaustible driving force for the prosperity of a country, research of science and technology and innovation are very important to a country for its development of science and technology. In the era of big data, researchers need to find useful information to do resreach from the huge volume of scientific literature. This will inevitably bring heavy workload. Generating a new and valuable research idea efficiently becomes a bottleneck in the research process. This paper proposes a text mining method for extracting a microscopic concept map(MCM) from a single scientific paper. This MCM visually describes the main knowledge structure of the author used in the paper. Then, through quantitative analysis of the concepts and their relationships in MCM, one can mine the author's research focus and the source of his new research ideas.This paper studies from the origin and development of concept map, and the constract and application of concept map, as well as the status quo of the research idea mining. Then we make further study of the concept and the relationship of the structure and the semantic relatio in the concept map, and propose a method for extracting dynamically a microscopic concept map(MCM) from a single scientific paper. Through quantitative description of the paper's literature research network provides help for the researchers of their science and technology research. The main contents include the following aspects. The effectiveness of the method is demonstrated by mining an example scientific paper.Firstly, we define and describe the microscopic concept map. This part introduces the composition and characteristics of microscopic concept map, and proposes two hypotheses.Secondly, we extract and construct the microscopic concept map. This construction process is mainly divided into two phases: One is the extraction rule setting of the MCM. The second is the constraction of the MCM of scientific papers according to the constraction rules. This stage mainly including the determination and pretreatment of target data source, the generation of the dependence tree of the simple sentence, the generation of the MCM fragments of the simple sentence, and the generation of MCM of the target data source.Thirdly, we mine the research ideas from the article base on the MCM. We refine the MCM, and describe it quantitatively, then make the weight and the distance of the main concepts of the MCM visually shown. We also use the thinking possibility index and the link prediction methods to mine the research idea.Finally, we have an empirical analysis case. In order to verify the validity of method this paper proposed, we select two articles published in Science about clustering method. Through the method for extracting a MCM that this paper proposed, we show the process of the MCM extraction and the mining process of the research ideas. We find the research idea that we have mined is the same as the author's research ideas. Therefore, this empirical case verifies the effectiveness of this method.