Abdelrahman M. Arab, Ahmed M. Gadallah and Akram Salah, Cairo University, Egypt
ABSTRACT
Classification is an important technique used in information retrieval. Supervised classification suffers from certain limitations concerning the collection and labeling of the training dataset. When facing MultiDomain classification, multiple training datasets and classifiers are needed which is relatively difficult. In this paper an unsupervised classification system is proposed that can manage the Multi-Domain classification problem as well. It is a multi-domain system where each domain represented by an ontology. A document is mapped on each ontology based on the weights of the mutual tokens between them with the help of fuzzy sets, resulting in a mapping degree of the document with each domain. An experiment carried out showing satisfying classification results with an improvement in the evaluation results of the proposed system compared to Apache Lucene.
KEYWORDS
Information Retrieval, Ontology, Machine Learning, Document Classification, Fuzzy Sets
Original Source URL: https://aircconline.com/ijsc/V8N1/8117ijsc01.pdf
https://airccse.org/journal/ijsc/current2017.html
No comments:
Post a Comment