Book Cataloguing System Using Optical Character Recognition and Named Entity Recognition

Abstract

Libraries function as knowledge preservers and agents of learning in communities. It is necessary to have access to well-organised books for efficient information retrieval and academic purposes. Nevertheless, traditional library systems’ cataloguing books are often manual and thus laborious and inefficient, making them prone to errors. This calls for an automatic system for cataloguing books to streamline the exercise, enhancing accuracy. Therefore, this research proposes an automated book cataloguing system that combines both Optical Character Recognition (OCR) and Named Entity Recognition (NER). In particular, the Tesseract OCR Engine is employed to grab text from book images while spaCy identifies and classifies entities existing within the document. All these functionalities are supported by a Python-developed back-end via an effective API, and the front end operates on React.js, ensuring user-friendly interaction. The result of this research reduces the time consumed to create catalogue records for books in the library, making the process efficient and easier. The system was evaluated based on Word Error Rate (WER) and Character Error Rate (CER) for OCR, while NER components were evaluated using precision, recall, and F1-score. The results of OCR have an overall of 3.2% for CER and 5.7% for WER. The year as a component under NER has the highest precision of 99%, while ORG has the lowest precision of 91%.