A similarity-based soft clustering algorithm for documents

Soft clustering algorithms based on neural networks

“…Frequency of particular term in particular document reflect how often the term appears in the document. All such document representations give Vector Space Model matrix X, where each document is represented as a column vector of this matrix (1) where is document vector and is the element of VSM matrix which gives frequency of j-th keyword in i-th document.…”

“…In our approach we used neural networks to soft clustering. It means, that each document can belong to the more than one category [1,2,4,11]. As training set we used the category representants, what are mixtures of documents that represent best each category.…”

A Methodology for Intelligent E-Mail Management

“…The clustering procedure is implemented by two connected and adapted algorithm. It uses a fuzzy hierarchical clustering algorithm to determine an initial clustering which is then refined using the SISC (King-Ip, 2001) clustering algorithm used in FISS meta-searcher structure.…”

Semi-supervised clustering techniques for categorization of text documents

“…It is noted that the term fuzzy clustering is sometimes mixed with the term "soft clustering" [43][44][45][46][47] in the data mining field. The research efforts made on developing soft clustering algorithms is very rich in literature.…”

“…in recent years, quite a few new similarity measures were explored under different DML algorithm, such as the Euclidean distance modified by a shortest-path algorithm [140], or Mahalanobis distances adjusted by convex optimization [141]. Many of them are focus on learning the family of Mahalanobis distances [29] which usually work well on low dimensional data, but are computationally expensive or even infeasible when handling high-dimensional data, or Bergman Distance [27,43,51]. Recently, Steven Ho [142] investigates how to learn a Bergman Distance function using a nonparametric approach.…”