If you have experience developing a recommender system, you are likely to have fallen victim to at least one of the following:
- The system is extremely slow when returning results due to the tremendous amount of datasets.
- Newly inserted data cannot be processed in real-time for search or query.
- Deployment of the recommender system is daunting.
So how can you build a product recommender system designed for massive amounts of data with ease? This article provides a solution to accelerating the candidate generation process by using MIND, PaddlePaddle & Milvus.
