Research On Knowledge Aggregation Of User Generated Content By Multi-Source Academic New Media

With the rapid development of information economy,academic new media,as a new online academic knowledge platform,has gradually attracted the attention of researchers,bringing changes to knowledge information acquisition,knowledge exchange and knowledge dissemination.Academic new media mainly exist in the forms of academic microblog,academic We Chat public account,academic virtual community and academic APP,which have the characteristics of diverse platform types,rich platform contents and professional platform knowledge.It provides a new way for researchers to obtain academic information,share academic achievements and carry out academic exchanges.Academic new media is no longer in the form of articles,journal articles and other long text as the content of knowledge dissemination.Academic users,as knowledge receivers and producers,coexist in the academic new media environment.The new media environment encourages academic users to generate new knowledge by asking,answering or sharing information,which innovates the way to acquire academic knowledge.With the expansion of the new media environment,the user-generated content in the Internet presents an explosive growth,and users need to spend a lot of time and energy to browse and filter the knowledge content in the process of searching knowledge.Knowledge content appears "knowledge overload" and the user is immersed in "knowledge trek".The user-generated knowledge content in academic new media also has some problems,such as uneven content quality,fragmented knowledge and redundant content.At the same time,there is a lack of information exchange between different academic new media platforms,and the knowledge in a single platform cannot be improved and updated in a timely manner.As a result,users have to spend a lot of time to browse the knowledge in multiple platforms,which increases the difficulty of acquiring knowledge.How to realize effective knowledge management,organization and mining of usergenerated content in academic new media,optimize and innovate knowledge service mode,and provide better knowledge service for users has become a new problem in the development of academic new media.In view of this,the knowledge aggregation theory and method are introduced into the knowledge aggregation of user generated content in academic new media.The thesis constructs the mechanism of multi-source academic new media user-generated content aggregation based on knowledge aggregation.The quality evaluation of user-generated content is discussed.Topic aggregation,summary generation and recommendation method are designed to solve the problems.Finally,this paper proposes a strategy to improve the ability of knowledge aggregation of usergenerated content in multi-source academic new media.This paper mainly carried out the following research:First,the knowledge aggregation mechanism of user-generated content of multisource academic new media is constructed.This paper defines the connotation of knowledge aggregation of the user-generated content in academic new media.It effectively organizes the knowledge contained in the user-generated content of new media platform,and then excavates the relation of its internal knowledge,so as to provide multi-source knowledge service for academic new media users.The types of knowledge aggregation of user-generated content in multi-source academic new media are divided into the following categories: homogenous knowledge aggregation,heterogenous knowledge aggregation,and multilingual knowledge aggregation.The knowledge aggregation and service elements of user-generated content in multi-source academic new media are divided into five aspects: knowledge service subject,knowledge service object,knowledge service content,knowledge service environment and knowledge service technology.This paper discusses the driving force of usergenerated content knowledge service in multi-source new academic media: the demand for academic information resources,academic innovation environment,revenue of knowledge service subject,scientific and technological progress,and multi-source academic resources.This paper expounds in detail the reasons of its influence on knowledge aggregation from the aspects of principle characteristics and mode of action.This paper explains the process of knowledge aggregation in multi-source academic new media,which includes user demand mining and interpretation,data processing of academic information,evaluation of academic information quality,knowledge discovery and aggregation,and academic knowledge recommendation.Finally,a model of knowledge aggregation mechanism for user-generated content in multi-source academic new media is proposed,and the specific process diagram of its practical application is drawn from the relationship and interaction among various elements.Secondly,an automated quality evaluation method for user-generated content in academic new media is proposed.The automatic feature extraction process of academic user-generated content is established by considering data dimension,emotion polarity and domain vocabulary.The double-layer BI-GRU neural network is used to learn the characteristic attributes of the academic user-generated content.The dictionary of professional academic domain is introduced to optimize the vectorized expression of academic user-generated text generated by word embedding model.Then the quality evaluation of user-generated content in academic new media is realized,and highquality text content is selected to provide a high-quality data basis for the follow-up research on knowledge aggregation method of user-generated content in multi-source academic new media.Thirdly,A knowledge subject aggregation method based on Bi LSTM-CNN-CRF and LDA models for user-generated content in academic new media is proposed.The hybrid neural network method is used to learn word segmentation of user-generated content in academic new media,and then the obtained academic professional word segmentation table is input into the LDA principal probability model to generate knowledge topics of multi-source academic new media.From the generated topics of multiple platforms,it is proved that there are great differences of knowledge topics among the same knowledge content of different academic new media platforms.On this basis,knowledge topics are integrated to help academic new media users to obtain core knowledge points from large-scale user-generated knowledge content and provide knowledge navigation services for academic new media users.Fourth,a method for generating knowledge abstractness of user-generated content of multi-source academic new media is proposed.Abstract generation method based on Word2 Vec model and MMR algorithm was proposed to realize the general knowledge description of user-generated content in multi-source academic new media.The Word2 Vec method can effectively solve the problem that the traditional method ignores the semantic relation between words.MMR sorting method is used to sort and screen the abstracts with high importance,and the abstracts with high repeatability are eliminated,while the abstracts with high importance are retained.The proposed method uses the domain dictionary to solve the problem that the domain words cannot be recognized generally.The aggregation of user-generated content on multiple academic new media platforms can help platforms with insufficient knowledge to make up for their own lack of knowledge.It realizes the great integration of the whole knowledge content in the academic new media environment.It provide an efficient aggregation service for academic new media users to acquire critical knowledge in redundant usergenerated content.Fifth,the thesis constructs an accurate recommendation method for knowledge aggregation of user-generated content in multi-source academic new media.The similarity relationship between the recommended items and users is mined through the user interest degree value,which is used as the user recommendation score for the items.Through the transfer of similarity,the similarity between users in multi-source academic new media platform is calculated,and then the user network is established.Finally,through the project recommendation score and user delivery network,the accurate recommendation method of knowledge aggregation of user-generated content in academic new media is constructed,so as to provide academic users with the dual accurate recommendation service of multi-source platform academic knowledge and academic users.