Classification of product description using machine learning algorithms

Product Classification Problem: ======================= There is Product Description Data (PDD) of several million records that includes products (such as yarn, diodes, eyeglasses, etc.). This PDD needs to be mapped to a product category (which includes a variety of products) Example Product Taxonomy Record: Wedding Dresses We need to classify these. Currently, we have an existing solution that uses simple text matching algorithms to classify. Based on that we are getting an accuracy rate of approx. 75%. Our objective is to get to 98% or higher. We are looking for experts who have actually solved this type of problem (in a professional setting and not just in a university) and can demonstrate how their solution would work with a sample of 5000 to 10000 records. Thanks Sumit SaS Technologies