Hierarchical classification of Web content

By Susan T. Dumais, Hao Chen. 2000

975 citations recorded by the source corpus.

This paper explores the use of hierarchical structure for classifying a large, heterogeneous collection of web content. The hierarchical structure is initially used to train different second-level classifiers. In the hierarchical case, a model is learned to distinguish a second-level category from other categories within the same top level. In the flat non-hierarchical case, a model distinguishes a second-level category from all other second-level categories. Scoring rules can further take advantage of the hierarchy by considering only second-level categories that exceed a threshold at the top level. We use support vector machine (SVM) classifiers, which have been shown to be efficient and effective for classification, but not previously explored in the context of hierarchical classification. We found small advantages in accuracy for hierarchical models over flat models. For the hierarchical approach, we found the same accuracy using a sequential Boolean decision rule and a multiplicative decision rule. Since the sequential approach is much more efficient, requiring only 14%-16% of the comparisons used in the other approaches, we find it to be a good choice for classifying text into large hierarchical structures.

Subfields: Artificial Intelligence, Information Systems

How researchers see this paper

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

In Artificial Intelligence

In Information Systems

Citations in the corpus

Large-Scale Hierarchical Text Classification with Recursively Regularized Deep Graph-CNN (2018, background): “Topical text classification is a fundamental text mining problem for many applications, such as news classification [18], question answering [28], search result organization [11], online advertising [2], etc.”
Machine learning in automated text categorization (2001, neutral): “An interesting evaluation has been carried out by Dumais et al. [1998], who have compared five different learning methods along three different dimensions, namely, effectiveness, training efficiency (i.”
PEBL: positive example based learning for Web page classification using SVM (2002, methodology): “As a binary classification algorithm, SVM gains increasing popularity because it has shown outstanding performance in many domains of classification problems [8, 10, 21].”
Hierarchical text classification and evaluation (2001, methodology): “A few hierarchical classification methods have been proposed recently [1, 2 , 10, 13, 15, 17].”
A Survey of Text Classification Algorithms (2012, methodology): “The SVM approach has also been used successfully [44] in the context of a hierarchical organization of the classes, as often occurs in web data.”
Hierarchical Classification via Orthogonal Transfer (2011, methodology): “One straightforward way for exploiting the hierarchical st ruc ure is to decouple the problem into a set of independent classification problems, each defined fo r an internal node in the hierarchy for classification between its immediate subclasses [6, 13, 4].”
Sentiment analysis: A combined approach (2009, background): “As explained in Dumais & Chen (2000) and Pang et al. (2002) given a category set, C = {+1, −1} and two pre-classified training sets, i.e., a positive sample set, T + r = n i=1 (d i , +1) and a negative sample set, T − r = n i=1 (d i , −1), the SVM finds a hyperplane that separates the two sets with…”
Support vector machines classification with a very large-scale taxonomy (2005, methodology): “The size of the training set for a category in a hierarchical SVM setup equals the number of documents in the sub-tree rooted by its parent category(4) [6].”
PEBL: Web page classification without negative examples (2004, methodology): “As a binary classification algorithm, SVM gains increasing popularity because it has shown outstanding performance in many domains of classification problems [7], [25], [26].”
Web site design with the patron in mind: A step-by-step guide for libraries (2006, methodology): “Various features have been used in Web page classification including words in the content of the Web page [7, 22], words associated with the links pointing to the page [9, 19, 22] and the neighbor Web page concept labels [4, 15, 16].”
User Profiles for Personalized Information Access (2007, background): “Recent approaches focus on extensions of traditional classification approaches to hierarchical concept hierarchies [23, 57].”
Bayesian models for Large-scale Hierarchical Classification (2012, background): “Some of the early works in HC [10, 14] use the hierarchical structure to decompose the classification problem into sub-problems recursively along the hierarchy and allocate a classifier at each node.”
Question classification using support vector machines (2003, background): “It would be interesting to extend the work using hierarchical SVM classifiers [6] in the future.”
Kernel-Based Learning of Hierarchical Multilabel Classification Models (2006, methodology): “In recent years, methods that utilize the hierarc y in learning the classification have been proposed by several authors (Koller and Sahami, 1997; McCallum et al., 1998; Dumais and Chen, 2000).”
Incremental Algorithms for Hierarchical Classification (2004, background): “Learning algorithms for hierarchical classification have been investigated in, e.g., [8] [9] [10] [11] [12] [14] [15] [17] [20].”
Web Search Intent Induction via Automatic Query Reformulation (2004, methodology): “Dumais et. al.”