Bouwmaat is a Dutch wholesaler that sells building materials to all kinds of organizations. They possess a wide product range with products like paint, building tools and many other products to make every task possible. With their typical slogan “You can build upon that” they are not only claiming to offer all products for builders, but also that they want to provide their customers with some proper advice.
Bouwmaat faced some difficulties regarding their product data. The data quantity appeared to be slightly too low and the quality of existing features appeared to be insufficient as well. This resulted in a challenge for the company: they wanted to provide their customers with sufficient information about the products, but in order to do that they needed data of more quantity and quality. That is why the company set a target: they wanted to have filled the missing product features by the end of the year. Bouwmaat had a hard time doing this without any assistance, so Squadra Machine Learning Company offered to help.
To see if this was achievable, a proof-of-concept was performed. To get this POC done, solutions from PowerEnrich.ai and PowerConvert.ai, software products from Squadra MLC, were combined. The so-called ‘web extraction’ function was used to extract data from websites to use this data in order to enrich the existing data from Bouwmaat.
After the proof-of-concept, it appeared that some of the data was useable, but some of the names and values of product features were completely different from Bouwmaat’s existing data. To solve this issue, a mapping needed to be conducted between Bouwmaat’s data structure and their suppliers’ data structures. By making use of Artificial Intelligence, Squadra MLC has proposed such a mapping in order to receive a better understanding of what a certain supplier text is about. However, this appeared to be more difficult than expected since the source data structure (Fest) wasn’t optimally connected to Bouwmaat’s data structure. Manual corrections were performed in order to fix this, and after that, everything appeared to be right- technically.
Product features were enriched, but it still appeared that the web extraction function couldn’t extract as much data as was expected. When it seemed like certain product features just weren’t able to be extracted, Squadra managed to find some solutions to connect the feature names from Fest to the attribute names from Bouwmaat. Due to the conflicting data structures, this required both AI solutions and manual solutions. As a result, both the data quality and quantity in the data set have improved.
With the help of Squadra MLC and its smart software solutions (PowerConvert.ai and PowerEnrich.ai), Bouwmaat has now managed to enrich its data with suppliers’ data. Since these suppliers use different data structures, it formed a challenge for the organization. Manually converting data would be an extremely time-consuming process that would cost way too much. The smart algorithms that Squadra MLC used, however, have helped to overcome this challenge in an efficient manner. In this way, valuable man-hours and unnecessary costs were saved.