In the expanding world of online retail, there exists an extensive catalog of products in the current retail markets. Different retailers and e-commerce sites have millions of product images and text descriptions. Matching the products across the universe proves to be an important and challenging task to determine if specific products exist in our catalog. Expanding upon our previous work on product matching and category mapping from textual descriptions using transformer based models, we propose a weighted multi-modal approach for product matching by using both images and text within the training and matching process. We integrate both transformer and ResNet architectures into the siamese network to generate fine-tuned product embedding. Extensive experiments are conducted to evaluate our proposed weighted multi-modal approach comparing with single-modal approaches. The experiments show our proposed approach outperforms single-modal approaches.

As the world's largest retailer, Walmart's core mission is to save people money so they can live better. We call the strategy we use to accomplish this goal our Every Day Low Price strategy. By keeping operational expenses as low as possible, we can continually apply a downward pressure on our prices, in turn increasing the amount of traffic, and ultimately, sales within our stores. In this paper, we apply Machine Learning (ML) algorithms and Operations Research techniques for forecasting and optimization to build a new price recommendation system, which improves our ability to generate price recommendations accurately and automatically. Comprised of a demand forecasting step, two optimizations, and causal inference analysis, our system was evaluated in the form of forecast backtests and live pricing experiments, both of which suggested that our approach was more effective than the current rule-based pricing system.