Research On Relevance Judgement Model Of Scientific Data Users

Driven by the open science environment and a "data-intensive" scientific research paradigm,the importance of scientific data continues to rise.Under the influence of FAIR principle and linked data technology,more scientific data enter the network for dissemination in the form of rich semantics.However,we know very little about how scientific data users make the judgment of whether a certain scientific data is relevant or not.Therefore,this paper focuses on the behavior and cognitive mechanism of scientific data users’ relevance judgment,understanding and modeling scientific data user relevance judgment.Based on this,it enriches and expands the relevance research in the field of information science,and at the same time provides the theoretical basis for the research and development of scientific data-specific search technology.Relevance is one of the basic concepts of information science.Understanding the process,principle,influencing factors and effect of users’ relevance judgment on specific types of information objects has always been the focus of information science,especially information retrieval scholars.Scholars have successively studied the relevance of various types of scientific papers,text documents,web pages,multimedia,social media,etc.Big data and scientific data are the latest concerns and attract more and more research interests due to their cognitive and technical characteristics are different from other information types.This paper attempts to build a user relevance judgment model for scientific data in an empirical way.Through the description of user relevance judgment process,quantitative verification of core cognitive process and algorithmic design of the model,the theoretical basis is provided for scientific data search and recommendation.For this reason,this study proposes a cognitive-oriented hypothetical conceptual model of relevance judgment and corresponding research objectives on the basis of summarizing the existing researches.Based on this,three levels of empirical research have been carried out in sequence: 1)Descriptive conceptual model research on user relevance judgment of scientific data;2)Empirical quantitative model research on user relevance judgment of scientific data;3)Multi-criteria relevance ranking(MCRR)algorithm and its application framework design.The following conclusions are drawn:Firstly,the paper verified the descriptive conceptual model of scientific data user relevance judgment through static element identification.In this model,24 scientific data information elements(SDIEs),3 types of decision rules,12 relevance criteria and 4 types of relevance criteria dimensions are identified.Based on this,the criteria structure of TAQAU(Topicality,Availibility,Quality,Authority and Usefulness)is proposed,which provides a "static" guiding principle for user-oriented scientific data query and retrieval practice.Secondly,based on PLS-SEM quantitative method,the paper empirically verified the relevance criteria using structure model in the interactive process of scientific data query and retrieval,namely the empirical quantitative model of relevance judgment.It verified the prerequisite function of topicality judgment in user relevance judgment,the intermediary function of availability,quality and authority,and the usefulness as the target variable for users to judge the relevance of scientific data.At the same time,the paper verified that the personalized cognitive mode of scientific data user relevance judgment is mainly reflected in the difference of dynamic combination of relevance criteria.Thirdly,based on the research findings and results of descriptive conceptual model and empirical quantitative model,the multi-criteria relevance ranking(MCRR)algorithm and its application framework are designed.The algorithm comprehensively considers the cognitive nature of users’ multi-criteria relevance judgment: 1)comprehensively calculates multi-criteria scores instead of the single dimension of objective relevance under the subject word matching mode;2)The priority operator is introduced to calculate the hierarchical relationship between different criteria(sets).At the same time,the research proposed a cognitive-oriented scientific data retrieval and recommendation system framework to support personalized scientific data retrieval and recommendation practices.According to the above research results and findings,the main academic contributions and innovations of the paper are as follows: 1)The paper innovatively proposes and validates the user relevance judgment model of scientific data from three levels(descriptive conceptual model/empirical quantitative model/algorithmic expression).At the same time,the paper extends the research object of user relevance judgment to scientific data and its users,enriching and developing the research process of user relevance in theory;2)Exploring the continuous research path of user relevance judgment from descriptive conceptual model to quantitative model and then to algorithm expression;3)A cognitive-oriented scientific data retrieval and recommendation framework is proposed,which provides a theoretical basis for intelligent and personalized scientific data retrieval and recommendation.Future research will further enrich and develop cognitive-oriented scientific data query and retrieval research in theory and practice.In theory,expand the research situation and types of users(e.g.scientific data experts in specific fields),and improve and develop the existing theoretical models;In practice,based on the theoretical model of user relevance judgment,the ultimate goal is to develop a cognitive-oriented scientific data query and recommendation system.So as to enhance the reuse and value of scientific data and support scientific research innovation under open science.