Identification And Empirical Study Of Content Features Of Domain Emerging Topics

Innovation is the foundation of the country's prosperity and the soul of national progress.As an important part of innovation,scientific research and innovation are of great significance to the overall construction of innovative countries and the economic transformation of scientific research results.Therefore,how to identify research and innovation activities is particularly important.Scientific research innovation is not produced out of thin air,but its footprint can be found.In the vast academic research,there are scientific innovations,which can be pursued by tracking emerging topics in academic literature,which is also the focus and difficulty of the field of knowledge management in Manage Science and Engineering.Studies on detection of emerging research topic will greatly enrich the field of knowledge management theoretically and pratically.This article focuses on the discovery of emerging research topics.Discovering emerging research topics using citation networks in the field,knowledge management,this paper finds that the essence of the methodology is to use citation networks to identify research topics and to screen out research topics in the emerging stage out of those identified topics.This kind of research idea has rationality and superiority,but it also destructively encapsulates the academic literature(the fulltext content)and the citation relationship between documents(citation content)and cannot exert their due values.With the increasing enrichment of the full-text data of academic literature and the maturity of natural language processing technology,the citation network can be added weights to leveraging from extracting full-text content features and citation content features to restore those encapsulated content.To better conduct this research,this paper first explores what content features can be used to discover research topics and then we apply these features to discover emerging research topics and evaluate our results.Along the way to identify content features,construct citation networks,identify research topics,screen emerging research topics,and validate research results,the main work of this paper includes the following four parts:(1)Exploring the relationship between citation content features and citing behavior in academic literature,and identifying citation content features that can be used for emerging research topic discovery.To better display the relationship between citation content and citations,this part of the study selects the citations of the papers published in the H index and their citances.Through the investigation of the relationship between the citation content characteristics and the time variation of the academic influence of the cited literature.This part of the study found that the citation content of the academic literature does have a good characterization of its academic influence,especially in the three characteristics of citation mention,average citation mention,and citation location.Therefore,the relevant citation content features identified in this part can be used as a discovery study for emerging research topics.(2)Exploring the distinguishing relationship between the content characteristics of the full text and the academic literature from the perspective of academic influence,and identifying the full-text content features that can be used for the discovery of emerging research topics.This part selects the academic papers of the two disciplines of biological science and psychology in PLo S and analyzes the relationship between the characteristics of the full text and its academic influence.The study extracted 12 features describing the full text of the academic paper through the CFA framework.By designing the null model,they are compared with the normalized citation frequency.The research in this part finds that under the existing full-text content characterization framework,using those full-text content features for the discovery of emerging research topics will have great interference,and it is impossible to use this type of content features to discover emerging research topics.The results of this part of the study initially indicate that the citation content characteristics of the two content features are more suitable for the discovery of emerging research topics.(3)Exploring the structural differences of the document coupling networks constructed via different content weighting strategies.This part compares the advantages and disadvantages of the full-text data of multiple full-text data providers and selects the biomedical field in PLo S as the data object for analysis.By combining the PLo S full-text data and the Wo S data,the document coupling network in the biomedical field is constructed and formulated.Eleven strategies(including unweighted)weight the edges of the network.By comparing the 11 coupling networks and discovering the related features such as citation content,the structure of the constructed document coupling network has changed in a certain amount.The weight distribution,degree distribution and node centrality of the nodes have significant changes.(4)An empirical study of emerging research topics combined with content characteristics.Based on the 11 coupled networks that have been constructed,this part identifies the research topics,constructs indicators for emerging research topic discovery,discovers emerging research topics,and analyzes and verifies the research results.The results show that the indicators constructed in this paper are useful to the discovery of emerging research topics;the citation location features have a significant effect on the discovery of emerging research topics;the citation mention feature has important value in the interpretation of research topics;the emerging topics show more advantages in long-term scientific impact than those non-emerging topics in our dataset.In general,this paper identifies some key problems in the research of emerging research topic discovery by combing the research and theory of emerging research topics and puts forward the research ideas of using the content characteristics of the literature to help the citation network to construct emerging research topics.This paper realizes the identification of the content characteristics of emerging research topics through two stages and four sub-studies,constructs the document coupling network by using the content features and discovers the emerging research topics,compares and analyzes the results of the traditional methods and the methods proposed in this paper,and verifies both results.This paper clarifies the research implications of this thesis from four aspects: 1 content features for emerging research topic discovery can improve the effect of discovery;2.content features to find emerging research topics have certain application value;3.content features could be applied to other similar research questions;4.the need to pay attention to its effectiveness when using content features,the adoption of inappropriate content features may also interfere with experimental results.