Loading...
Please wait, while we are loading the content...
Similar Documents
Dtesign t and evaluation of clustering approaches for large document collections, the “bic‐means ” method by (2006).
| Content Provider | CiteSeerX |
|---|---|
| Researcher | Hourdakis, Nikolaos |
| Abstract | The availability of large volumes of digital content in modern applications (e.g., digital libraries and organization intranets) and the on internet has generated additional interest in methods and tools for effective management of shared content. Data clustering is a means for achieving better organization of the information by partitioning the data space into groups of entities with similar content. Clustering of large document collections is the problem this thesis is dealing with. State of the art clustering algorithms are reviewed first (e.g. partitional and hierarchical algorithms). Initially, we focus on partitional clustering methods due to their low time complexity (i.e., linear on the number of documents). Hierarchical clustering methods are considered as well. We examine several variants of the original K-Means algorithm and we propose the so-called “Incremental K-Means ” which differs from K-Means in the way the centroids are updated during each clustering iteration. However, both K-Means and its variants produce a flat partition of the data collection. Efficient methods which are able to provide effective organization of information |
| File Format | |
| Publisher Date | 2006-01-01 |
| Access Restriction | Open |
| Subject Keyword | Bic Mean Method Several Variant Digital Content Data Clustering Flat Partition Low Time Complexity Partitional Clustering Method Hierarchical Algorithm Organization Intranet Additional Interest Efficient Method Digital Library Hierarchical Clustering Method Data Collection Data Space So-called Incremental K-means Effective Organization Modern Application Effective Management Large Volume Similar Content Original K-means Algorithm Large Document Collection |
| Content Type | Text |
| Resource Type | Thesis |
