A Product Information Management (PIM) system is essential for manufacturers, wholesalers and retailers in supporting the sales process by passing through product information. Due to the emergence of artificial intelligence, it is possible to make these PIM systems ‘smarter’ and thus obtain better product data for more products in less time.
Machine learning is the technique for artificial intelligence, that enables software to learn without begin explicitly programmed. Instead of writing code with a lot of “if this then that” rules, it recognizes patterns in data and creates an algorithm that can recognize, predict, advise, optimize and/or classify. It is an extremely powerful technique that allows software to perform intelligent tasks like:
- Classification: predicting the correct product classification for new products;
- Feature matching and value conversion: Automatic matching of product features and values of supplier product data with standards (e.g. ETIM/GS1) or company structures to greatly reduce manpower;
- Product Enrichment: Enriching product data using algorithms on images, scraping manufacturers websites and scraping of PDF product documentation;
- Text generation: generating product descriptions and/or marketing texts and translating them into other languages;
- Variant detection: analyzing product datasets and detecting variants for these products, e.g. replacements, duplicate items, alternatives, products with slightly different unit of measures, colours, dimensions, etc. Or cluster products in product groups.
Squadra machine learning company wrote a white paper which explains how to transform a PIM system into an intelligent PIM solution and what the benefits are.
If you are interested, please download the whitepaper.