Some details
Large electronic component distributor company approached us with a problem of automating their online purchase workflow.
Their customers are used to submit the lists with descriptions of products to be ordered in freely written form that has to be linked with a real product by human.
We have developed Natural Language Processing (NLP) model that does Named Entity Recognition (NER) over customer search lines and
identifies various kinds of product parameters, such as electrical units of measure and their relevant values, part numbers, dimensions, case codes e.t.c.The recognized product parameters then were used for relevant product search. The developed model has achieved ~90% accuracy in identifying electronic parts parameters.