<![CDATA[Named Entity Recognition (NER) is a core task in natural language processing for extracting structured entity information from text. However, funding descriptions on peer-to-peer lending platforms are largely unstructured. This limits reliable identification of funding categories and location-related entities required for subsequent analysis. This study addresses this problem by applying Named Entity Recognition to identify agricultural entities from peer-to-peer lending funding descriptions in Indonesia, combining it with distance analysis. The data used in this study was collected from peer-to-peer lending platforms in Indonesia using web crawling techniques with the Python Selenium library. The collected funding data was used to train and test a Named Entity Recognition model developed using the spaCy library, with entity labeling performed using the Beginning–Inside–Outside (BEIN–IND–OUTS) tagging scheme. Model performance was evaluated using a confusion matrix at the token level. The evaluation results showed that the proposed model achieved 83% accuracy, 94% precision, 82.7% recall, and a 90% F1 score, indicating its ability to detect agricultural entities from lighting descriptions. Furthermore, the collected data containing agricultural entities was processed using the Haversine formula to calculate the distance between the lender and the borrower's location. When compared to Google Maps, the average distance difference was 23.7 kilometers. These results demonstrate that Named Entity Recognition combined with distance analysis can support the preparation of peer-to-peer lending data for further decision-making.]]>
Copyrights © 2025
