Text Classification: An Implementation Project

1
Text ClassificationAn Implementation Project

• Prerak Sanghvi
• Computer Science and Engineering Department
• State University of New York at Buffalo

2
Algorithm to be used

• I intend to use a BackPropagation Artificial
  Neural Network
• Inputs are in terms of whether a particular
  keyword is present or not in a document
• Output is in terms of the category into which the
  document should be classified

3
What are the keywords?

• This falls under a broader class of problems,
  known as Feature Selection.
• Some technique in Feature Selection will be used
  to automatically or semi-automatically pick the
  keywords.

4
Organization of the Project

• The project will really consist of two phases,
  each of which is equally important for good
  results
• Feature Selection
• Implementation of the ANN

5
Artificial Neural Network
6
Example data set

• One of the several corpora available on the web
  will be used

7
After ANN

• Once the technique to extract the feature set
  from the data set is implemented, any algorithm
  can be used to make the classification.
• After ANN is successfully implemented, other
  algorithms, especially Naïve Bayes classification
  method will be implemented.
• Comparison of results from different methods will
  be compared.
• Another possibility is the coupling of two
  methods to improve the overall performance.