Utilizing TensorFlow for Information Extraction from Arabic Commercial Documents | ||||
| Delta Journal of Science | ||||
| Articles in Press, Accepted Manuscript, Available Online from 24 August 2025 | ||||
| Document Type: Research and Reference | ||||
| DOI: 10.21608/djs.2025.399214.1220 | ||||
 View on SCiNiTO | ||||
| Authors | ||||
omnia elbarbary   1; Shaimaa Hagras2
| ||||
| 1faculty of computer science and informatics, Tanta university | ||||
| 2Information System Department, Faculty of Computer and Informatics, Tanta University, Egypt. | ||||
| Abstract | ||||
| The vast amount of data contained within commercial documents presents a challenge for businesses seeking to unlock valuable insights. This paper explores the use of TensorFlow, a powerful open-source machine learning framework, for building an automated information extraction system specifically designed for commercial documents. The proposed system leverages deep learning techniques to identify and extract relevant information from diverse document types, including contracts, invoices, purchase orders, and financial reports. We discuss the limitations of traditional manuals and rule-based information extraction methods. This paper details the data preparation process for commercial documents, the deep learning model architecture built with TensorFlow, and the training and evaluation methodology employed to assess the system's performance. The results demonstrate the effectiveness of the proposed approach in accurately extracting crucial data points from various types of commercial documents. This research contributes to the field of document information extraction and offers a valuable tool for businesses to automate data extraction tasks, improve operational efficiency, and gain deeper insights from their commercial documents. | ||||
| Keywords | ||||
| Information Extraction; Arabic Text Processing; Natural Language Processing (NLP); Deep Learning; TensorFlow | ||||
|
Statistics Article View: 2 |
||||
