Kaemingk is a leading importer of distributor of seasonal decorative items. With a team of around 450 employees Kaemingk presents the most innovative home deco collections for Christmas, spring/summer, Easter and Valentine, each year. 

The product range consists of more than 20.000 decoration items for home and garden that are displayed in showrooms in the Netherlands, Belgium, Germany and the United States. Deliveries are made to thousands of professional customers in over 80 countries.

Challenge 

As part of the renewal of the IT application landscape, Kaemingk has implemented a new PIM system. It was a requirement to enter the products that Kaemingk has in its range in a structured manner into the new PIM system. The existing data per product was contained in various textual descriptions which had to be converted into a list of individual product characteristics per product. Converting this data manually is a very time consuming process, and it also requires the available time of employees who are very busy and also needed for other processes.

Solution

With the help of smart algorithms, the extraction of the product characteristics is largely automated. This has resulted in a significant reduction in manual conversion work and a consistent data set.

Part of the conversion had already been done manually by the product specialists. Based on this data, algorithms have been trained to determine which keywords in the text are associated with each feature. These keywords are further expanded with the knowledge of the product specialists so that the correct conversions are recognized by the software.

/var/folders/fn/zjq5ch890b36qfpr0mkr3j_c0000gn/T/com.microsoft.Word/Content.MSO/C246CB59.tmp

Result

The conversion of the Christmas 2019 and Spring 2020 datasets resulted in ~90% of the found feature values to be confident enough to approve directly. To reach full coverage we built a webapplication in which the product specialists could check and complete the products in their own domain. Both the application and the algorithm have been designed in cooperation with these specialists to ensure quality and ease of use. 

 
Want to know more? 

For more information or a demo, contact Guus van de Mond: gvdmond@machine-learning.company / +31 (0)629 502018.  

Guus van de Mond

"The development and application of innovative solutions is the strength of our team. I would like to get in touch with you to discuss the possibilities for your organization.''