Original link: tecdat.cn/?p=984

 

Exploration of intelligent recommendation engine in e-commerce industry

Machine learning helps maternal and child e-commerce

The profile

Tuoduan helps domestic maternal and child e-commerce companies to create intelligent recommendation engines, so as to create accurate and efficient shopping experience, explore how to implement various recommendation strategies on large-scale data, optimize strategies, and build efficient recommendation engine solutions.

Business challenges

With the increase of the number of users and goods on e-commerce websites, data becomes an important factor affecting recommendation quality. As a hot area in e-commerce, the value of trillions of China on maternal and infant market with two child policy full liberalisation has entered the era of high growth growth, maternal and child consumer market each year, more than 30 billion maternal and infant consumption can be added at least bring an annual 13% new growth space, huge market inevitably contains huge business opportunities and great profit space.

As we all know, the main ways to solve information overload include category navigation, search, recommendation, and the current hot chatbot, but its essence is also based on recommendation system and knowledge graph. Recommendation is different from or better than search: search requires users to know what they need, while recommendation can help users discover what they need or let the information you need find you, and it is more personalized, and can even “know yourself better than yourself”.

The traditional recommendation mechanism mainly includes the working principle of recommendation mechanism based on demography and the basic principle of content-based recommendation mechanism.

How demographics-based recommendation mechanisms work

Douban’s recommendation “Douban chai”

The basic principles of content-based recommendation mechanisms

Meanwhile, due to the fierce competition in the mother-baby market, the homogenization of products is becoming more and more serious, and the traditional recommendation mechanism can hardly meet the business needs.

For this cooperation, the main challenge is how to design an intelligent recommendation engine to accurately find the products that users need from the mass of products.