Abstract:
The current book classification system in libraries usually assign only one category to each book, while the content of a book pertains to a variety of categories. The Internet classification also not comply with any international standard, and has no related between the Internet information and the books in library.
This research were presented Automated Dewey Decimals Multiple Relation Classification System consisted of the Information Extraction, the Keyword Analysis, the evaluation of Keyword Weight, the categorization are accordance with the DDC-MR Rules. The first stage is the automatic classification for specific category we compared the classified category with information from OhioLINK. The efficiency of our prototype are measure Accuracy, Precision, Recall and F-Measure. The second innovation is the automatic DDC-MR classification, showing the category pertaining to the book content and was also analyze article title in Wikipedia, which allow to link to the content in the library.
This research of experiment results could separate into 2 groups : (1) Only use DDC-MR Rule the efficiency are Accuracy 88 %, Precision 90%, Recall 96 % and F-Measure 93 % and (2) use Neural networks on efficiency are Accuracy 73 %, Precision 78%, Recall 73 % and F-Measure 71 %. The other results from this work is the analysis of information could be compare values books in the library to classified as the same groups of information.