Open-Set Classification for Automated Genre Identification

Abstract

Abstract. Automated Genre Identification (AGI) of web pages is a problem of increasing importance since web genre (e.g. blog, news, eshops, etc.) information can enhance modern Information Retrieval (IR) systems. The state-of-the-art in this field considers AGI as a closed-set classification problem where a variety of web page representation and machine learning models have intensively studied. In this paper, we study AGI as an open-set classification problem which better formulates the real world conditions of exploiting AGI in practice. Focusing on the use of content information, different text representation methods (words and character n-grams) are tested. Moreover, two classification methods are examined, one-class SVM learners, used as a baseline, and an ensemble of classifiers based on random feature subspacing, originally proposed for author identification. It is demonstrated that very high precision can be achieved in open-set AGI while recall remains relatively high.

Exploiting link structure for web page genre identification

“…To evaluate the performance of MCC, in addition to using the common majority voting (MV) [22] algorithm as the baseline method for comparison, we also implemented the Random Feature Subspacing Ensemble (RFSE) algorithm based on the latest work [31] without considering the case of an unknown genre to suit our purpose. We used the same features in all three methods.…”

“…Pritsos and Stamatatos [31] focused on the use of content information with di↵erent text representation methods. In addition, they examined base SVM learners and an ensemble of classifiers.…”

“…SVM Contents is based on the textual information on the page as described in Section 3.1 while SVM On-Page Features adds structural information in the feature set. SVMs based on textual and structural information have been used by various researchers for genre identification, such as in [13,31,34], as discussed in Section 2. SVM Neighboring Pages is based on the textual and structural information in the selected neighboring pages.…”

“…For example, a website like www.news.com.au, has many pages in di↵erent topics, e.g., sport news and finance news, but the genre of these pages in this website is news. Genre identification has recently attracted increasing attention because it can be used to significantly improve the quality of web and document search results [1,4,8,31,37]. With genre identification, a search engine can label the genre of each retrieved result for the users to save their time, as exemplified in Fig.…”

Term Weighting Based on Index of Genre for Web Page Genre Classification

“…One of the solutions to this problem is to classify web pages according to the genre of web pages [2]. Automating the identification of the genre of web pages has become an important area in the classification of web pages [3], as it can be used to improve the quality of web search results and to reduce the search time [4].…”