信息网络安全
信息網絡安全
신식망락안전
Netinfo Security
2015年
10期
86-91
,共6页
微博%用户关系强度%距离模型%加权有向图
微博%用戶關繫彊度%距離模型%加權有嚮圖
미박%용호관계강도%거리모형%가권유향도
microblog%user srelationship strengths%distance%directed weighted graph
对微博用户关系强度的评估是分析和研究微博网络的基础。文章提取微博用户的个人基本信息、交互信息、互粉信息，并提出一种基于距离的评估模型对微博用户的关系强度进行准确的评估分类，该距离模型根据三种假设，得出最基础的用户关系强度群，然后根据距离函数对未知的关系强度进行分类。由此得出一张有向加权图，节点表示微博用户，边表示用户之间的关注关系，权值表示微博用户之间的关系强度。将该距离模型得出的预测结果和采用对用户相对交流值，用户背景的相似度和用户是否互粉三种属性值进行加权融合得出的实验结果进行比较分析可得出：该评估模型能够结合用户的综合信息评估出用户之间的关系强度，结果全面、准确、直观，也充分说明该模型既吸取了加权融合法的优点，也避免了加权融合的弊端，很好的体现了机器学习的优势，为微博舆论的研判提供最直接的参考依据。
對微博用戶關繫彊度的評估是分析和研究微博網絡的基礎。文章提取微博用戶的箇人基本信息、交互信息、互粉信息，併提齣一種基于距離的評估模型對微博用戶的關繫彊度進行準確的評估分類，該距離模型根據三種假設，得齣最基礎的用戶關繫彊度群，然後根據距離函數對未知的關繫彊度進行分類。由此得齣一張有嚮加權圖，節點錶示微博用戶，邊錶示用戶之間的關註關繫，權值錶示微博用戶之間的關繫彊度。將該距離模型得齣的預測結果和採用對用戶相對交流值，用戶揹景的相似度和用戶是否互粉三種屬性值進行加權融閤得齣的實驗結果進行比較分析可得齣：該評估模型能夠結閤用戶的綜閤信息評估齣用戶之間的關繫彊度，結果全麵、準確、直觀，也充分說明該模型既吸取瞭加權融閤法的優點，也避免瞭加權融閤的弊耑，很好的體現瞭機器學習的優勢，為微博輿論的研判提供最直接的參攷依據。
대미박용호관계강도적평고시분석화연구미박망락적기출。문장제취미박용호적개인기본신식、교호신식、호분신식，병제출일충기우거리적평고모형대미박용호적관계강도진행준학적평고분류，해거리모형근거삼충가설，득출최기출적용호관계강도군，연후근거거리함수대미지적관계강도진행분류。유차득출일장유향가권도，절점표시미박용호，변표시용호지간적관주관계，권치표시미박용호지간적관계강도。장해거리모형득출적예측결과화채용대용호상대교류치，용호배경적상사도화용호시부호분삼충속성치진행가권융합득출적실험결과진행비교분석가득출：해평고모형능구결합용호적종합신식평고출용호지간적관계강도，결과전면、준학、직관，야충분설명해모형기흡취료가권융합법적우점，야피면료가권융합적폐단，흔호적체현료궤기학습적우세，위미박여론적연판제공최직접적삼고의거。
Evaluation of user relationship strengths of Microblog is a basis for analysis and research of Microblog network. This paper fetches personal information, mutual information and fans information of Microblog users, puts up with a distance-based evaluation model to accurately evaluate and classify the relationship strengths between Microblog users. The model based on three assumptions, draw up the most basic user relationship strengths group, and then the relationship between the intensity of the unknown are classified according to the distance function. Compared the predictions derived from the model and result which derived through three attribute value weighted fusion: the relative exchange value of the user, the user background similarity and whether the usermutual followed, we can infer that with comprehensive, accurate and visible results, the model can evaluate the relationship strengths between users combined with their general information, andindicates that this model both combines the advantages of weighted fusion method and avoids the disadvantages of the weighted fusion, which indicated the advantages of machine learning, which provides the most direct references for studying and judging Microblog comments.
