Multi-models Dynamic Recommendation for Offline Clothing Sales

In the era of data-drive-productivity, the effective use of data will lead to higher conversion rates and turnover for offline clothing sales. This paper proposes a dynamic recommendation algorithm that combines multi-models such as collaborative filtering, content-based recommendation, visual-based recommendation, and hot sales models. According to the purchase records in the different period of a certain season, the weights in different recommendation models were dynamically adjusted to get a better recommendation result. Experiments proved our algorithm would achieve an excellent recommend performance, the generated results were rich and complex, the recommended results for new customers were as smart as for regular customers.