A Comparison of Document Clustering Techniques

By Michael S. Steinbach, George Karypis, Vipin Kumar. 2000

3,316 citations recorded by the source corpus.

Subfields: Natural Language Processing, Model Architectures

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Natural Language Processing

In Model Architectures

Citations in the corpus

Visual Genome: Connecting Language and Vision Using Crowdsourced Dense Image Annotations (2016, methodology): “We use hierarchical agglomerative clustering (Steinbach et al. 2000) on vector representations of each”
Comprehensive survey on hierarchical clustering algorithms and the recent developments (2022, methodology): “Bisecting k-Means is a divisive hierarchical clustering method (Steinbach et al. 2000).”
TwitterStand: news in tweets (2009, methodology): “We then use a variant of the cosine similarity measure [17] for computing the distance between t”
Hierarchical Clustering (2017, methodology): “ctive. For a deﬁnition and overview of agglomerative algorithms (such as average-linkage, complete-linkage, and single-linkage) see e.g., (Friedman et al., 2001) and for divisive algorithms see e.g., Steinbach et al. (2000). Most previous theoretical work on hierarchical clustering aimed at evaluati”
Frequent term-based text clustering (2002, methodology): “A lot of different text clustering algorithms have been proposed in the literature, including Scatter/Gather [2], SuffixTree Clustering [3] and bisecting k-means [4].”
Unsupervised Identification of Translationese (2015, methodology): “The described method for selecting a clustering outcome can be viewed as a binary version of the Bisecting KMeans algorithm; it is applied in all experiments throughout the paper, with number of iterations ( N ) ﬁxed to 5, following the recommendation by Steinbach et al. (2000, p. 13).”
Efficient phrase-based document indexing for Web document clustering (2004, methodology): “Steinbach et al. [32] give a good comparison of document clustering techniques.”
Top 10 algorithms in data mining (2007, methodology): “Bisecting k-means [73] achieves this by first putting all the data into a single cluster, and then recursively splitting the least compact cluster into two using 2-means.”
Data Mining in Healthcare and Biomedicine: A Survey of the Literature (2012, background): “Several research studies have showed that Bisecting K-mean, a variation of K-mean, provides the best clustering accuracy 4 [73, 79 ].”
Automatic online news issue construction in web environment (2008, methodology): “This result does not agree with the conclusion of [11], which finds out that “agglomerative hierarchal clustering performs poorly” and explains that “in many cases, the nearest neighbors of a document are of different classes”.”
Introduction to Information Retrieval (2008, neutral): “An example of an efficient divisive algorithm is bisecting K-means (Steinbach et al. 2000).”
Similarity Measures for Text Document Clustering (2008, background): “As mentioned previously, partitional clustering algorithms have been recognized to be better suited for handling large document datasets than hierarchical ones, due to their relatively low computational requirements [16, 9, 3].”
Correlation clustering in general weighted graphs (2006, neutral): “[8, 14, 19, 20, 21, 25].”
A Survey of Clustering Data Mining Techniques (2006, methodology): “2L There is experimental evidence that compared with Forgy’s algorithm, the second (classic) version frequently yields better results [Larsen & Aone 1999; Steinbach et al. 2000].”
Exploiting Wikipedia as external knowledge for document clustering (2009, background): “4.3 Evaluation Metrics Cluster quality is evaluated by three metrics, purity [14], F-score [10], and normalized mutual information (NMI) [15].”
Recommender Systems for Large-scale E-Commerce : Scalable Neighborhood Formation Using Clustering (2002, background): “This algorithm is fast and tends to produce clusters of relatively uniform size, which results in good cluster quality [14].”