Product/Brand extraction from WikiPedia

TL;DR

This paper presents a method for extracting product and brand pages from Wikipedia using a probabilistic classification approach, demonstrating promising results and discussing alternative methods.

Contribution

It introduces a novel probabilistic classification method for identifying product and brand pages in Wikipedia, along with an experimental setup and dataset.

Findings

The probabilistic model achieved promising accuracy in classifying product pages.

The experimental environment facilitates future research in Wikipedia page classification.

Alternative approaches were considered but the probabilistic method showed competitive results.

Abstract

In this paper we describe the task of extracting product and brand pages from wikipedia. We present an experimental environment and setup built on top of a dataset of wikipedia pages we collected. We introduce a method for recognition of product pages modelled as a boolean probabilistic classification task. We show that this approach can lead to promising results and we discuss alternative approaches we considered.