社会计算中基于人格特征的用户建模

发布时间：2022-12-10 00:44

　　随着社会计算系统的蓬勃发展,越来越多的信息和特征被用于用户建模,如画像信息、位置、行为和偏好等。社交媒体为分析用户情绪、个性等内在状态提供了各种各样的资源。用户的个性特征作为一种有价值的资源,可以反应被研究用户的内在特点,这启发了一项新的研究领域,即个性计算。现阶段,该领域的研究大部分集中在通过分析用户数据自动识别用户个性,很少将用户个性特征纳入到推荐系统中,更没有研究用户的个性特征对用户建模、兴趣挖掘过程以及推荐准确性的影响。本文提出了一种基于Big Five人格模型和用户兴趣动态建模的个性化感知用户建模框架。为了证明该框架的高效性,我们设计了以下三个应用场景:（1）提出了一种新颖的基于Big Five个性特征模型和混合过滤的朋友推荐系统。依靠个性特征和用户和谐度实现推荐过程,实现了名为PersonNet的社交网站,以此证明该推荐系统的准确率。（2）设计了一种基于动态主题建模和Big Five个性特征的用户兴趣挖掘系统。为了验证在兴趣挖掘过程中融入用户个性特征的高效性,构建了一个支持新闻共享的社交网络,并对收集到的数据进行了不同的实验。（3）构建了一种基于用户兴趣挖掘和元路径发现的个... 

【文章页数】：103 页

【学位级别】：博士

【文章目录】：
Acknowledgements
摘要
Abstract
Chapter Ⅰ Introduction
    1.1 Social computing
    1.2 Personality traits theory
    1.3 Personality computing
    1.4 Recommendation systems
    1.5 User interest mining
    1.6 Problem statement and research questions
    1.7 Innovations and contributions
        1.7.1 Personality-aware friend recommendation system
        1.7.2 Personality-aware user interest mining system
        1.7.3 Personality-aware product recommendation system
    1.8 Thesis structure
Chapter Ⅱ Related works
    2.1 Automatic personality recognition
        2.1.1 Text-based APR
        2.1.2 Image-based APR
        2.1.3 Gaming and Behavior-based APR
    2.2 Personality enabled social robots
    2.3 Personality in recommendation systems
        2.3.1 Friend recommendations
        2.3.2 Multimedia recommendations
        2.3.3 Academic content recommendations
        2.3.4 Product recommendations
    2.4 User interest mining
Chapter Ⅲ PersoNet: Friend Recommendation System Based on Big FivePersonality Traits and Hybrid Filtering
    3.1 Introduction
    3.2 Notations
    3.3 System model
    3.4 Similarity measurement
    3.5 Recommendation system
    3.6 Experiment details
        3.6.1 Data
        3.6.2 Participants
        3.6.3 Personality measurement
        3.6.4 Data collection phase
        3.6.5 Harmony rating
        3.6.6 Friend recommendations
        3.6.7 Testing phase
    3.7 Performance evaluation
        3.7.1 Implementation
        3.7.2 Evaluation metrics
        3.7.3 Results discussion
    3.8 Conclusions
Chapter Ⅳ: Mining User Interest Based on Personality-aware Hybrid Filtering inSocial Networks
    4.1 Introduction
    4.2 Notations
    4.3 Representation model
        4.3.1 User modeling
        4.3.2 Topic modeling
        4.3.3 Implicit interest prediction
    4.4 System evaluation
        4.4.1 Dataset and experiment details
        4.4.2 Variants
        4.4.3 Baselines
        4.4.4 Evaluation metrics
        4.4.5 Results analysis and discussion
    4.5 Conclusions
Chapter Ⅴ: Personality-aware Product Recommendation System based on InterestMining and Meta-path Discovery
    5.1 Introduction
    5.2 Notations
    5.3 System design
    5.4 Representational model
        5.4.1 Users representation
        5.4.2 Topics representation
        5.4.3 Items representation
    5.5 Interest mining
    5.6 Item mapping
    5.7 Meta path discovery
    5.8 Evaluation
        5.8.1 Baselines
        5.8.2 Evaluation metrics
        5.8.3 Dataset description
    5.9 Results discussion
    5.10 Conclusions
Chapter Ⅵ: Conclusion and future directions
References
作者简历及在学研究成果
学位论文数据集



本文编号：3715707