นงนุช จินดารัตนาภรณ์ 
Expected outcome | Details and deliverables |
| Short term outcomes (1st year) | Phase 1 - An update report for Search Module and Detection Module implementation o A web crawler collects data from search engines (e.g., Google, Bing). The data collected should include text, links, and images of the product. o An Optical Character Recognition (OCR) program extracts text from images. Phase 2 - An update report for Text Classification Module implementation o A machine learning or deep learning model detects illegal food advertisements, and drug-like property advertisements. Phase 3 - AI system for classifying potential illegal food advertisement. o The system uses keywords and scans for food advertisement on the internet. o The system can detect text from advertisement images. o The system can cross-check with FDA database and classify an illegal advertisement. o The system can create reports containing details of potentially illegal advertisements, classification o The system can create reports containing details of potentially illegal advertisements, classification results, confidence scores, and relevant FDA information. - A final report and feasibility study on the system |
| Middle term outcomes (2nd year) | Tool's real-time detection capabilities will allow Thai FAD
to respond promptly to breaches of existing
law, thereby preventing longer-term harm and will provide evidence for strengthening regulatory frameworks. |
| Long term outcomes (2nd year) | Advantages extend to the broader adaptability of the tool to other areas of digital marketing such as tobacco, alcohol, and breastmilk substitute (BMS). |