Research On Text Mining Of Academic Innovation Knowledge For Academic Evaluation

Academic evaluation can measure the academic research role,influence and contribution of researchers,and at the same time can identify the academic innovation carried by the research results.It is an essential part of building an innovative country.Using scientific and reasonable methods to objectively and fairly evaluate academic papers’ innovation has always been a key research issue for information researchers.The evaluation methods for the innovation of academic papers are mainly divided into two categories: the qualitative evaluation method with peer evaluation as the primary method and the quantitative evaluation method based on bibliometrics.However,both methods have certain limitations,and neither can meet the academic evaluation requirements of the new era of science and technology strategy.On the one hand,the traditional qualitative method based on peer review is relatively reliable under ideal conditions.However,the efficiency is not enough in actual operation,and it is easily affected by non-academic factors.On the other hand,quantitative methods represented by SCI and other quantitative indicators have specific reference value,but most of the existing indicators are external descriptions of evaluation objects,ignoring academic innovation’s internal evaluation.At present,indicators such as SCI index,impact factors,and citations have become the core for evaluating scholars’ innovation ability and contribution.This evaluation thinking that uses quantitative data to evaluate academic contributions has seriously affected Chinese academic ecology.Under the pursuit of SCI indicators in the academic world,academic misconducts such as plagiarism and academic fraud have emerged one after another,which has had a negative impact on the level of scientific development in China,resulting in more and more calls for improving academic evaluation methods.This article is based on text mining theory and technology,combined with comprehensive information theory and other theories to research academic innovation knowledge mining.We propose a complete set of academic innovation knowledge mining methods and propose application framework and function design ideas.We try to help the identification,extraction and evaluation of core innovations in academic research in order to improve the work efficiency of evaluation experts,reduce their burdens,and better assist evaluation experts in carrying out quantitative and qualitative academic innovation evaluation work,which provide choices for the academic evaluation method in the new era.Based on the analysis of related research about academic evaluation and text mining,this research focuses on the following aspects:(1)A method for mining academic innovation points for academic evaluation is proposed.Based on discussing the current mainstream mining methods,this research proposes an extraction method that combines the deep learning model Bert and rules to construct the method and process of mining innovation points in academic texts based on the characteristics of automatic identification elements of academic innovation.This method uses manual labeling of innovation points,training the best Bert classification model,and formulating extraction rules to achieve the mining of innovation points.Finally,it is verified by empirical research to determine the feasibility of the method.(2)The identification method of the academic innovation domain for academic evaluation is proposed.Based on the successful mining of academic innovation points,this paper proposes a process and method for identifying academic innovation domains based on the Topic2 Vec model to measure the similarity between research innovation topics.Through text preprocessing,LDA topic recognition,Topic2 Vec semantic mining,semantic mining result verification,and other steps,we achieve the identification of academic innovation topics and use case studies to test this method’s feasibility.(3)A calculation method of academic innovation value for academic evaluation is proposed.To obtain potential knowledge from academic texts that reveal the innovation level of academic literature,this research introduces the concept of academic semantic measurement into the field of Chinese academic texts for the first time.Based on the analysis of Chinese academic texts’ features,an optimized calculation process of academic innovation values is proposed.Through data set construction,text preprocessing,vectorized representation,innovation value calculation and result verification,an optimization algorithm based on Doc2 Vec is proposed.Finally,the method’s feasibility is verified based on examples,and the measurement of academic innovation value is achieved.(3)A framework for the application of academic innovation knowledge for academic evaluation is designed.This research refers to the classic process of text mining to design the academic innovation knowledge application framework’s main body: knowledge storage layer,knowledge mining layer,and knowledge application layer.On this basis,the application framework’s construction ideas are further proposed,which are the acquisition,cleaning,labeling of academic data,and the mining and application of academic innovation knowledge.Specific applications include academic innovation time series,academic innovation theme maps,and academic innovation field maps.The application framework designed in this article is expected to assist the efficient and low-cost implementation of academic evaluation and stimulate users’ desire for academic exploration,enhance users’ ability to capture academic innovations,and thus facilitate the practice,innovation,and development of scientific research.Based on relevant theories and methods such as comprehensive information theory and text mining,this paper proposes the concept of academic innovation knowledge mining,expands academic evaluation based on the perspective of text mining,and enriches the means for scientific innovation and academic evaluation.At the theoretical level,this research puts forward the concept of academic innovation knowledge mining.It integrates cutting-edge information technologies such as machine learning and text mining into the process and methods of innovative knowledge mining in academic texts,enriching and perfecting the theory and method of academic innovation knowledge.At the practical level,under the guidance of the concept of academic innovation knowledge mining,through empirical research,the processes of academic innovation point mining,innovation domain identification and innovation value calculation are respectively discussed,and on this basis,the main composition and construction ideas of the application framework are analyzed.This research has obtained the academic innovation knowledge mining process that can be used as a reference for the academic community,promotes the exploration of academic innovation evaluation,helps to improve the efficiency of scientific evaluation and the discovery speed of academic innovation,and further promotes the achievement of science and technology.In the future,an academic innovation knowledge application platform will be established to achieve the automatic mining of innovation points,innovation domains and innovation values in academic texts in order to promote the organic integration and development of text mining technology and academic evaluation.