WEB PAGE INFORMATION EXTRACTION – SMART SCRAPER
Problem
In order to extract information for any website, e-commerce companies need to develop a manual scraper for each website separately.
Solution
Creating smart scraper – a deep learning model which extracts the title, price, image, description and availability from any website for any product.
Results
The model extracted the content from any website with the following result:
- Extracting the images with 98.7% accuracy
- Extracting the price with 68% accuracy
- Extracting the name of the product with 72% accuracy
- Extracting the description with 60% accuracy
- Extracting the availability with 53% accuracy
These results were shown on a test data of 20.000 samples in 4 different languages: French, Italian, English, and German.