Supplier Recommendation Tool

Challenge

Corporate procurement is tasked with selecting suppliers and purchasing various goods and services. To consistently achieve the best price-performance ratio, comparing suppliers and obtaining competitive offers are essential. Our client, an international automotive corporation, sought to generate a list of alternative supplier suggestions for any chosen reference supplier.

Approach

First, the data from existing offer documents had to be prepared for modeling. Available service descriptions were thoroughly analyzed and cleaned. Filler words were deleted, meaningful words were reduced to their stems, and normalized according to their rarity and significance. Then, using Latent Semantic Analysis (LSA), the suppliers were positioned in a semantic space based on their service descriptions. By translating textual information into a numerical space, numerous mathematical and logical operations can be applied to the suppliers or offer documents. For example, similarities can be calculated to display relevant supplier recommendations.

Result

Our NLP model generates suggestions for corresponding suppliers—the nearest neighbors in the semantic space—starting from a reference supplier. The model was integrated into an interactive dashboard with a user interface for generating suggestions and facilitating searches in the supplier database. The application simplifies the use of existing information, reduces work time, and saves our client money by enhancing supplier competition in each tendering process.