SMU Data Science Review


As the internet age evolves, the volume of content hosted on the Web is rapidly expanding. With this ever-expanding content, the capability to accurately categorize web pages is a current challenge to serve many use cases. This paper proposes a variation in the approach to text preprocessing pipeline whereby noun phrase extraction is performed first followed by lemmatization, contraction expansion, removing special characters, removing extra white space, lower casing, and removal of stop words. The first step of noun phrase extraction is aimed at reducing the set of terms to those that best describe what the web pages are about to improve the categorization capabilities of the model. Separately, a text preprocessing using keyword extraction is evaluated. In addition to the text preprocessing techniques mentioned, feature reduction techniques are applied to optimize model performance. Several modeling techniques are examined using these two approaches and are compared to a baseline model. The baseline model is a Support Vector Machine with linear kernel and is based on text preprocessing and feature reduction techniques that do not include noun phrase extraction or keyword extraction and uses stemming rather than lemmatization. The recommended SVM One-Versus-One model based on noun phrase extraction and lemmatization during text preprocessing shows an accuracy improvement over the baseline model of nearly 1% and a 5-fold reduction in misclassification of web pages as undesirable categories.

Creative Commons License

Creative Commons Attribution-Noncommercial 4.0 License
This work is licensed under a Creative Commons Attribution-Noncommercial 4.0 License

Included in

Data Science Commons