尋求合作者是科研工作者的重要學(xué)術(shù)活動之一,因?yàn)楹线m的合作者會有助于提高學(xué)者的研究質(zhì)量,加快其研究進(jìn)程。隨著信息技術(shù)的快速發(fā)展,科研社交平臺已經(jīng)廣泛出現(xiàn),并且吸引了大量的研究人員通過虛擬社區(qū)來開展科研合作。因此,在這些科研在線平臺中開發(fā)出高效的合作者推薦系統(tǒng)將有效地促進(jìn)學(xué)術(shù)合作與知識共享。信息過載和信息不對稱是在合作者推薦的研究領(lǐng)域需要解決的兩個(gè)關(guān)鍵問題�？偟膩碚f,本研究首先要定義出潛在的學(xué)術(shù)合作者推薦的情境,同時(shí)需要給出對應(yīng)的解決方案,為用戶提供有效的建議和決策支持�，F(xiàn)有的合作者推薦研究主要關(guān)注研究者之間的相似度,如基于專業(yè)知識背景的相似性和社交網(wǎng)絡(luò)的鄰近度等。盡管在這一領(lǐng)域已經(jīng)有很多的研究,但是對于科研合作者推薦的總體框架和有效的推薦算法仍然是缺乏的。在本研究中,我們提出了一個(gè)總體框架來解決科研合作者的推薦問題。該框架定義出了兩個(gè)主要的合作者推薦情境,即基于相似性的合作者推薦,和在一個(gè)特定的背景限制下的合作者推薦。針對這兩個(gè)推薦情境,本文提出了兩個(gè)對應(yīng)的高效的解決方案。對于基于相似性的合作者推薦問題,我們提出了一個(gè)混合方法,分別從專業(yè)知識的相關(guān)性、社交網(wǎng)絡(luò)的鄰近度和機(jī)構(gòu)層面的合作度三... 

【文章來源】：中國科學(xué)技術(shù)大學(xué)安徽省 211工程院校 985工程院校

【文章頁數(shù)】：109 頁

【學(xué)位級別】：博士

【文章目錄】：
摘要
ABSTRACT
1 INTRODUCTION
    1.1 Background and Motivation
    1.2 Research Objectives
    1.3 Research Approach
    1.4 Research Organization
2 LITERATURE REVIEW
    2.1 Academic Collaboration
    2.2 Collaborator Recommendation Approaches
        2.2.1 Content-based collaborator recommendations approaches
        2.2.2 Homogeneous network-based collaborator recommendations approaches
        2.2.3 Heterogeneous network-based collaborator recommendations approaches
    2.3 Semantic Analysis Techniques
        2.3.1 Probabilistic latent semantic indexing
        2.3.2 Latent Dirichlet Allocation
        2.3.3 Author-topic model
    2.4 Social Network Analysis Techniques
        2.4.1 Neighborhood-based network proximity measures
        2.4.2 Path-based network proximity measures
        2.4.3 Topology-based network proximity measures
    2.5 Context-constrained Recommendation Approaches
    2.6 Summary and Research Gaps
3 PROPOSED APPROACH
    3.1 A General Collaborator Recommendation Framework
    3.2 Similarity-based Collaborator Recommendation
        3.2.1 The similarity-based collaborator recommendation framework
        3.2.2 Expertise relevance
        3.2.3 Social network proximity
        3.2.4 Institutional connectivity
        3.2.5 Fusing strategy
    3.3 Collaborator Recommendation within a Specific Context
        3.3.1 Recommendation framework
        3.3.2 Expert quality profiling
        3.3.3 Quality-weighted LDA model
        3.3.4 Expertise coverage-oriented matching
4 EXPERIMENTAL EVALUATIONS
    4.1 Evaluation of Similarity-based Collaborator Recommendation
        4.1.1 Dataset preparation
        4.1.2 Experimental design
        4.1.3 Evaluation metrics
        4.1.4 Results and discussions
    4.2 Evaluation of Collaborator Recommendation within a Specific Context
        4.2.1 Dataset construction
        4.2.2 Experimental design
        4.2.3 Evaluation metrics
        4.2.4 Results and discussions
    4.3 System Implementation of the Two Proposed Approaches
        4.3.1 System implementation for similarity-based collaborator recommendation
        4.3.2 System implementation for collaborator recommendation within a specific context
5 CONCLUSIONS AND FUTURE WORK
    5.1 Summary of the Research
    5.2 Contributions of the Research
    5.3 Limitations of the Research
    5.4 Future Research
REFERENCES
致謝
在讀期間發(fā)表的學(xué)術(shù)論文與取得的其他研究成果



本文編號：3519661