الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 32 |  |  |
| | | from | 32 | | |
Abstract
Over the past decades a considerable amount of time and effort has been expended researching and developing web mining techniques capable of extracting rules from data published on the web. The purpose of this thesis is to set out an investigation and develop a new automated web text mining system based on swarm technique that empower the rule extraction process for improving the teaching process. The thesis presents a generalized ant colony algorithm suitable for problems of a combinatorial nature. Specifically, if one can find an efficient algorithm (i.e., an algorithm that will be guaranteed to find the optimal solution in a polynomial number of steps) for one problem, then efficient algorithms could be found for all other problems. Many ant colony algorithms developed and tested to solve many problems of different nature .we proposed a modified algorithm to solve any problem of combinatorial nature like Travelling sales man and data mining. The proposed algorithm is successfully implemented and tested to solve travelling sales man problem. The results are compared with the original algorithm. The experimental results prove that the proposed algorithm using the new stopping condition can save the time and the CPU Usage. ACO Principles applied to data mining classification problem. Our experiments has been performed on the database from the famous WEKA system in such a way that the results of the ant algorithm is compared to the results of the famous WEKA system algorithms according to a specific criteria such as ; accuracy rate and rule simplicity. The proposed algorithm approved to be the best choice for data mining application compared with other related famous data mining algorithms. In such a way, its results in mapped as a higher accuracy rate and simpler rule generated. Finally, the implemented generalized ACO for data mining algorithm used to develop a web text mining application that handles the drawbacks within the training process in an educational center (Orascom training technology (OTT)) as a case study. The experimental results prove that the extracted rules improve the classification process that enhance the available feedback information and then the overall evaluation process.