Client: An e-commerce enterprise connecting small stores to customers using their platform
Problem: The client has thousands of products that are hosted on their service. Many products are not properly tagged, and some are not tagged at all by the seller. This makes it hard for the customer to find the product they are looking for and this leads to lost opportunities. Client needs a software solution that automates the tagging and metadata generation process. Client is also expecting an increase in the number of sellers and visitors and needs to be ready for expansion into different segments due to the new sellers on their service. The problem can be broken down as
- Setup a system that automates the tagging process and that generates metadata for each product, allowing the service to find products in fast, easy manner for better customer experience.
- Deploy models that will be able to understand the differences and similarities between different products through user rating, description, and search queries.
VT’s Solution: We deployed advanced analytical models and utilized Artificial neural networks to process the data collected by the client. By utilizing the search queries, the system was able to self-determine what type of classification the users were most frequently looking for. Learning the different segments of products and defining the ways in which people looking for such a product distinguish them. By doing so, the AI system was able to slowly determine the products tags and metadata based on the description, features, and search queries. To allow for new products and product segments, we trained the AI system with various other product datasets. Now as the system is self-learning, it will automate the scaling as new sellers join the service and would also create tags and metadata based off the ways in which people search for the product. This improved the clients service, by removing the inefficiencies and creating a better customer experience.
Key Points:
- Achieved client’s business objective by creating a system that automates tagging and generates metadata for each product.
- Designed and implemented an Artificial Neural Network to allow the system to continuously learn from the different search queries and descriptions, providing ways to segment the products that are inline with the ways most people search
- Provided a system that was scalable and able to handle new types of data, trained the AI with different product data to allow easy transition into new product segments.