Studies On Research Front Detection Under The Context Of Time Distribution Characteristics Of Key Words

Research front detection,as an important part of scientometrics,is of great significance to the academic field.The research front of a subject field often represents the most important and core research theme and development trend in the field,and it has high reference value for research topics selection and national science and technology policies formulation.Since the concept of the research front was proposed,many scholars have advanced different methods.The mainstream methods include word frequency analysis method,Burst terms detection method,co-citation analysis method and co-word analysis method.It' s of great importance to detect research front under the context of utilizing key words,keyword-based analysis methods usually use the keyword word frequency as the raw data for research front detection.Based on the size of the word frequency and how fast the word frequency changes,the development trend of the research and theme evolutionary processes is analyzed and the front of research in the subject area is detected.The word frequency method is widely used because of simple data processing and intuitive result analysis.However,the existing word frequency methods are not perfect.In the existing methods,in order to reflect the change process of the word frequency in the time dimension,the word frequency is usually processed by time slicing.This operation simplifies the data calculation,but ignores the time distribution characteristics of the keywords in the time window,so that the keyword differences within the same time window disappear.And the continuity of word frequency changes throughout the time axis is also destroyed.The purpose of this study is to come up with a new method that can more accurately characterize the evolution of keywords on the basis of retaining the original distribution characteristics of keywords more comprehensively.In this study,all high-frequency keywords included in articles of the seven core disciplines of statistics,accounting,film and television arts,surveying and mapping,dental science,plant protection,and safety science during 2008 to 2018 recorded in Summary of Chinese Core Journals are used as empirical data.The data in the time range is fitted to the cumulative distribution function of the keywords,and the word frequency cumulative speed and word frequency cumulative acceleration are used to characterize the heat and potential of keywords in the field.On this basis,the two dimensions of heat and potential are used to detect the research front and analyze the dynamic evolution.The data processing and analysis of this study mainly focused on three levels.At the keyword level,the dynamic evolution of high-frequency keywords can be converted into two continuous functions of word frequency cumulative speed and word frequency cumulative acceleration.The development status and trend of keywords at each moment can be expressed by the index of heat and potential.At the level of the subject area,the high-frequency keywords are used to draw a heat-potential distribution map based on the heat and potential value at the same time,and then the keywords are divided into key front keywords,high potential keywords,high heat keywords and general keywords according to the position of the keywords on the distribution map.At the multi-disciplinary level,this study conducts a correlation analysis on the multi-year heat and potential rankings of keywords in 7 subject areas,and compares the results of correlation calculations between different disciplines.The results show that although the 7 disciplines are developing,the specific development process is different for different disciplines.In addition,this study also compares the detection results under the heat-potentiality index with detection results under the existing word frequency method.The comparison results show that there is a difference between the detection results under the heat-potentiality index and the detection results under the existing word frequency method.Therefore,the application of the time window in the word frequency method is likely to bring deviation to the research front detection results.The detection method that retains the time distribution characteristics of keywords is a useful exploration trial for the temporal cumulative evaluation.However,this study is not yet perfect,and the fitting of the time distribution of keywords still needs more attempts and further research.