Abstract: This paper analyzes the deficiencies of the FTSVM. To overcome these shortcomings, it proposes a latent link analysis-based FTSVM(LLA_FTSVM) for hypertext classification. The new method takes the information of weights of hypertexts into consideration. Experimental results show that the LLA_FTSVM performs well

Key words: Hypertext classification; TSVM; Fuzzy; Latent link analysis; Weight of hypertext

摘要： 回顾了模糊直推式模糊支持向量机（FTSVM）的不足，并提出了一种基于潜在链接分析并结合网页权重信息的FTSVM 网页分类方法。新方法提高了分析网页超链接信息的效率，避免了经验参数的影响，充分考虑了网页权重的贡献。试验表明，提出的方法在网页基准测试数据上取得了优于FTSVM 的分类效果。

关键词: 网页分类；直推式支持向量机；模糊；潜在链接分析；网页权重