الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 144 |  |  |
| | | from | 144 | | |
Abstract
This study depends on the technique of artificial intelligence and introduces model as administrative system of the library based on multi-agent to help librarians in classifying books and periodicals automatically. It consists of two agents namely; Book or Periodical Classification and Returning borrowed Book or Periodical. The first agent consists of several stages, the first stage is the stage of preprocessing & feature extraction and this stage is divided into several steps, after capturing the cover image of the book or periodical using the camera sensor, the title to be categorized is defined and converted into text using the Ocrade java script function, the Tokenization step is initiated and the title is divided into a list of words and symbols, followed by the Removing stop step, to delete stop words and extract unique words, The Stemming word step is then performed to reduce some words to their base. The features vector is then obtained and the iterations are calculated using TF-IDF Algorithm (Term Frequency – Inverse Document Frequency).Finally, the KNN classifier calculates the distance between each feature from the input vectors and all the words in the terms data base and find the minimum distance and store the class who it belong to. Finally, it will be N suggested classes; the most repeating class will be the final identification matching result. The domain name category to which the book or periodical belongs is displayed, as well as the classification number. Then the classification agent takes an action to save the book or periodical data within the books and periodicals database. Also the second agent consists of several stages, This is the image acquisition phase, followed by the preprocessing phase, and finally the KNN classifier to calculate the similarity between the bar code of the book or the periodical returned with the barcodes of the books stored in the database of borrowed books and periodicals, then an event is taken action updates that database after deleting the book or periodical returned. The proposed system also contains two databases: The first terminology database includes categories of scientific departments in the Faculty of Specific Education, Mansoura University (Computer Science - Home Economics - Music Education - Art Education - Educational Media) and most of the words related to each category, The second: the database of books and periodicals borrowed. In order to evaluate the performance of the proposed system, , Recall, F-measure and accuracy were used. The experimental results confirmed that the proposed algorithm provides good classification efficiency. In general, the study results demonstrate the effectiveness of the proposed system. With the help of the proposed system, librarian will be able to manage library and classify books and periodicals with high accuracy and efficiency.