Hybrid Neural Network and Rule-Induction Approach for the Subjective Cataloging and Classification of Document. | ||||
| MEJ- Mansoura Engineering Journal | ||||
| Article 3, Volume 25, Issue 1, March 2000, Page 1-19 PDF (363.35 K) | ||||
| Document Type: Research Studies | ||||
| DOI: 10.21608/bfemu.2021.146261 | ||||
 View on SCiNiTO | ||||
| Author | ||||
| A. E. El-Alfy* | ||||
| Specific Education Faculty., El-Mansoura University., Mansoura., Egypt. | ||||
| Abstract | ||||
| The subjective cataloging process of researches and books depends on the experience of the classifier. Although the index terms given by the authors at the end of their abstracts can guide the classifier to the proper subject; they are not quite enough to express the real content of the research. The Title and the abstract of a given research play an important role in the subjective cataloging. This paper utilizes the human index terms given in the papers published in the leading journals to build domain thesaurus Tries (advanced B-Tree). The Tries has the possibility to locate the index term and its occurrence. A rule induction system is used for the subjective cataloging be extracting the effective features index terms) from the title and abstract of a given research. The domain thesaurus' Tries and the rule induction system are used to classify new document by supervised artificial neural network (SANN). The training mode of the SANN is enhanced by three main algorithms, the genetic algorithm (GA), the conjugate gradient algorithm (CGA) and the simulated annealing algorithm (SAA). The processes of training and testing the SANN in the document classification are also presented. | ||||
| Keywords | ||||
| Domain thesaurus Tries; B-Tree; Rule induction system; Supervised neural network; Genetic algorithm; Conjugate gradient; Simulated annealing algorithm | ||||
|
Statistics Article View: 73 PDF Download: 142 |
||||
