Research On Academic Hot Spots And Achievements Personalized Recommendation System In Textile

With the explosive growth of the number of papers in the past two decades,the academic community has entered the era of big data.The paper recommendation system filters the recommendation results according to the published papers of scholars,and recommends the papers according to the actual needs of scholars.The academic hotspot mining technology is based on the theory of bibliometrics to help scholars track the changes of research hotspots.Both can reduce the burden of academic information collection and combing of researchers.At present,the bottleneck of the application of academic hot spot mining technology in the textile field is that there is no unified collection,disambiguation and classification of professional vocabulary(research direction)in the textile field,which leads to inaccurate hot spot mining.The bottleneck of the application of scientific and technological information service platform in the textile field is that it is difficult to disambiguate scholars in the whole subject field.At present,the identity of scholars is only identified according to their names and institutions,and the papers published by textile scholars under the signature of different institutions are not completely collected,which leads to the inaccurate modeling of scholars and affects the recommendation of papers.This topic applies paper recommendation and academic hotspot mining technology to the textile field to provide personalized,private and diversified paper recommendation services for scholars in the textile field.The main research contents are as follows :(1)Based on the collection of 36 Chinese journals of《China Academic Journal Impact Factor Annual Report(2022 Edition)》published by China National Knowledge Infrastructure(CNKI),and the work of more than 190,000 textile Chinese papers published until July 10,2022,this study constructed a textile scholar database(6294scholars),disambiguated the homonymous scholars in the textile field based on the co-author relationship,and used LDA,Doc2 Vec and Node2 Vec models to extract the characteristics of scholars’ research fields,paper semantics and co-author relationship,so as to automatically draw digital portraits for textile scholars.The time change of scholars’ research direction is visualized by the theme river map.(2)Based on the collection of《Textile Chinese Thesaurus》《Textile Encyclopedia》《Textile Dictionary》and 199861 key words of academic papers,this study uses the Chinese-English correspondence of author keywords to achieve word sense disambiguation,and constructs a Chinese textile terminology database(17806 terms).Combined with the bibliometric characteristics of terms(time,citation and co-occurrence relationship),the relationship between research technology and materials is excavated,and the future academic hotspots are predicted to provide a basis for paper recommendation based on academic hotspots.(3)Based on the construction of textile terminology database,paper database and scholar database,this paper develops an academic personalized recommendation system in the textile field.The system can recommend papers related to the research direction of scholars in recent years,papers published by scholars with similar research directions,and papers published by scholars with close academic relations to scholars in the scholar database.At the same time,taking into account the academic needs of junior researchers in the textile field(fewer published papers,lack of sufficient information to model scholars),the system is recommended to their academic hot papers.Compared with the mainstream recommendation systems CNKI and AMiner,the academic recommendation system developed in this project has the following advantages :(1)In the process of constructing the data base of textile scholars,this system disambiguates the scholars in the field of textile,collects the papers published by different institutions,and draws the digital portrait of textile scholars.In the process,the opinions of textile experts are integrated,so that the computer can more reasonably identify the changes in the research interest of textile scholars.(2)When mining academic hotspots in the textile field,this system collects common terms in the textile academic field.On the basis of disambiguation and classification,the main material and technical terms in each branch of textile field are sorted out,which provides basic data for the application of academic hotspot mining technology in the whole field of textile and improves the accuracy of hotspot mining.(3)When recommending papers for textile scholars,this system takes into account the personalized and diversified needs of scholars,recommends papers related to the research field of scholars,reduces the burden of scholars’ retrieval,recommends papers published by scholars closely related to scholars’ academics,helps scholars understand the dynamics around them,recommends papers published by scholars related to the research field of scholars,helps scholars understand their peers,recommends academic hot papers,and helps scholars grasp frontier hot spots.This system has initially realized the digitization of the textile academic field.It is not comprehensive enough to collect only Chinese papers published by scholars and model them.It only recommends Chinese papers to scholars and cannot meet the needs of scholars for English papers.At present,the overall framework of the system has been basically completed,and the automation of data collection,cleaning,disambiguation,training and paper recommendation process has been realized.The current bottleneck is the lack of two functional modules: cross-language homonymous disambiguation and machine translation for the textile field.When these two functions are realized,the system can better meet the needs of scholars in the textile field.