A summary of Using SVMs For Text Categorization

Susan Dumais et al.; DOI

For the summary of the paper, go to the Summary section of this article.

Table of Contents

• A summary of Using SVMs For Text Categorization
  • Table of Contents
  • First Pass
    • Category
    • Context
    • Contributions
  • Second Pass
    • Background Work
    • Motivation
    • Figures, Diagrams, Illustrations, and Graphs
    • Clarity
    • Relevant Work
    • Future Directions
  • Summary
  • Summarization Technique
  • Citations

First Pass

Discussion about the title, abstract, introduction, section and sub-section headings, and conclusion

The essay Using SVMs For Text Categorization by Susan Dumais et al. (as part of the larger Support vector machine collection of essays in the July/ August edition of the 1998 IEEE Intelligent Systems magazine) [1] provides examples of when using a Support Vector Machine (SVM) is beneficial with respect to text classification. They discuss text classification, text representation and feature selection, and an example use case on the Reuters collection. They support the position that using SVMs for text classification (or really any algorithm so long as it isn’t run by a human) is beneficial for this task.

Category

What type of paper is this work?

This essay is more argumentative and position oriented.

Context

What other types of papers is the work related to?

This essay would most likely be classified alongside similar works that evaluated the usefulness of SVMs with respect to human tasks.

Contributions

What are the author’s main contributions?

Their contributions is an analysis of SVMs for text classification.

Second Pass

Background Work

What has been done prior to this paper?

Work has been done to understand what SVMs are, as well as use cases for SVMs.

Motivation

Why should we care about this paper?

Because it provides a case study of using SVMs on the Reuters collection with respect to text classification.

Figures, Diagrams, Illustrations, and Graphs

Are the axes properly labeled? Are results shown with error bars, so that conclusions are statistically significant?

All of the graphs and charts are clear to understand and have properly labeled axis.

Clarity

Is the paper well written?

This essay is clearly written.

Relevant Work

Mark relevant work for review

The following relevant work can be found in the Citations section of this article.

• Introduction to Modern Information Retrieval [2]
• Fast Training of SVMs Using Sequential Minimal Optimization [3]

Future Directions

My own proposed future directions for the work

I would like to implement their study using the five different learning algorithms they utilized to validate their results. The algorithms in question are: Findsim, Naive Bayes, BayesNets, Trees, and LinearSVM.

Summary

A summary of the paper

The essay Using SVMs For Text Categorization by Susan Dumais et al. (as part of the larger Support vector machine collection of essays in the July/ August edition of the 1998 IEEE Intelligent Systems magazine) [1] presents the usage of SVMs for text categorization on the Reuters collection in comparison to other classification algorithms. They found that SVMs perform best on this classification task.

The greater reason for this essay is to encourage engineers to use learning algorithms for human intensive tasks - such as text classification.

Summarization Technique

This paper was summarized using a modified technique proposed by S. Keshav in his work How to Read a Paper [0].

Citations