Abstract: Behind this in the end is our self-control is not enough, or the e-commerce platform is too to read people’s hearts, we might as well from the technical dimension, to explore.

This article is shared from Huawei Cloud community “618 Technology Special (1) Unconsciously over budget 3 times, why can’t you stop buying?” Torchbearer of technology.

On June 18, you can’t help but feel like you’re being pulled by some magical force that once you’ve added something to your shopping cart, you can’t stop.

Watching live with goods is also the same, I heard the anchor Shouting in the ear: buy one and then send one, buy two and then send three, hand on the order.

Whether we lack self-control or e-commerce platforms are too good at reading people’s minds, we might as well explore the details from the technical dimension.

Graph database: Establish relationships between entities and gain insight into your preferences

Xiao Wang, a new father from Ju Factory, plans to make a good show of himself on 618 by buying milk powder bottles for his baby. Xiao Wang opened an e-commerce APP and added the milk powder to the shopping cart. Then he slid it down and saw that the toy was good and the suit of clothes was cool. The operation was as fierce as a tiger, and wang became more and more excited by the purchases, ending up buying three times more than he had planned.

Yes, during the promotion period, Xiao Wang inadvertently fell into the recommendation network of the e-commerce system. APP home page, shopping cart page, product details page… Recommendation systems are everywhere. The background of e-commerce APP has prepared a series of products for Xiao Wang based on the user’s portrait (gender, age, shopping history, search history, etc.).

In the field of e-commerce, recommendation system plays a powerful role, which enables users to spend more time browsing goods and achieve the purpose of increasing the unit price of customers. It is the graph database technology that is applied behind it.

People who are new to a graph database can easily be misled by its literal meaning into thinking that it is a database for storing images, which it is not. Just as Lei Feng and Leifeng pagoda are two completely different concepts, graph database refers to an online database management system that stores data in a graph structure.

For example, Ju Chang Xiao Wang and Xiao Li are colleagues and both like to play table tennis. Here, Xiao Wang and Xiao Li are entities respectively, representing a point. The relationship between Xiao Wang and Xiao Li — colleagues, is the edge connecting the two points, which is a simple graph structure.

Through the graph structure, we can model all kinds of related scenes, ranging from the recommendation system of social network and e-commerce platform to the transportation system of the whole city.

For example, the e-commerce platform will label Xiao Wang according to his characteristics (programmer, dad, fond of playing table tennis, etc.) and judge the user attributes with the labels: Xiao Wang usually likes to shop for digital products of BRAND A and is used to buying baby products of brand B. However, Xiao Li, who also often reads the mobile phone information of brand A, will not buy the products of brand B. With the help of the graph structure, this association can be found out to achieve accurate marketing.

If you want the graph structure to find this relationship, you have to rely on the query analysis, calculation, storage management, visualization, etc. For example, graph database Neo4j is good at real-time query of graph data. The graph engine focuses on off-line analysis and mining of massive graph data using mature graph algorithms.

Huawei Cloud Graph Engine Service (GES) provides query and analysis services for relations-based graph structure data. Taking Huawei Mall as an example, it can realize real-time product recommendation by means of GES. It analyzes and compares the preferences of target users and other users, and recommends the products purchased by other users to target users after finding similarities.

One of the “heroes” of GES is EYWA, which provides a complete solution from low-level graph storage and management, core HIGH-PERFORMANCE computing engine, to upper-level graph analysis and graph query.

In terms of technology, EYWA made these optimizations:

  • Distributed optimization ParallelSliding Window(PSW) graph computing framework, efficient loading graph data to meet the needs of large-scale business computing;

  • Considering the efficiency of graph calculation and point query, the block data organization based on edge set is developed to organize data reasonably.

  • The prefetch strategy of edge set is used to hide disk I/O operation and the relaxed BSP model is used to hide communication I/O to improve performance.

Another major “hero” is the graph scene and graph optimization algorithm owned by GES. Take the Pixie algorithm as an example. Pixie is an algorithm designed by Huawei Cloud to construct multiple data into the same graph and configure corresponding schema, point-edge attributes and weights on this heterogeneous graph. As a new real-time recommendation algorithm, it overcomes the problem of data acquisition and fusion of heterogeneous graphs, supports comprehensive recommendation under multi-request nodes, and can meet the requirements of various complex, time-varying and diversified recommendation scenarios.

Therefore, the real-time recommendation algorithm provided by GES graph engine service makes recommendations with higher accuracy under the joint action of multiple relationships (historical interaction information between users and commodities, potential relationships between people and commodities, etc.). With a large amount of data, it can still achieve a good real-time recommendation effect and has strong scalability. For specific data, please refer to this article: Core Algorithm Of Huawei Cloud New Generation black Technology Revealed.

Knowledge map: Michelin Chef “processing” data raw materials, intelligent recommendation fast, accurate, ruthless

If graph databases emphasize the storage, query and management of “data raw materials,” the Knowledge graph is the Chef of the Michelin restaurant, processing data further. The knowledge graph based on graph engine service fuses all kinds of heterogeneous and heterogeneous data to form a large-scale knowledge base to support business applications and make search results more accurate.

Based on graph database, knowledge map extends the correlation properties of all commodities (goods of various dimensions, these dimension properties, goods to the amount of social evaluation, etc.), due to extend the hidden relationship between users and commodities, supplement the user interactions with the item data, based on people – people, realize people – things. Therefore, the recommendation effect can be further improved.

For example, users with the same attributes may be interested in the same kind of goods. When Xiao Wang, who likes playing table tennis, buys the quick-drying clothes of brand A on the e-commerce APP, then xiao Li, his colleague who also likes table tennis, opens the APP, the store that Xiao Wang just bought may be in front of him. You see, this recommendation system, based on acquaintance attributes, has been quietly automated through the knowledge graph.

On the one hand, the construction of knowledge graph improves the accuracy of personalized recommendation of e-commerce platform, and it can also be applied to intelligent customer service to help establish knowledge system cards and realize intelligent question and answer.

For example, when you enter “display” in the customer service dialog box of an e-commerce enterprise, the knowledge card will list the product introduction, features (screen size, pixels, etc.), style classification, suitable for people, suitable for occasions, production process, etc.

Huawei mall online customer service is also a typical application, it can deal with consumers in the pre-sale and after-sales consultation of all kinds of problems, timely reply to the product promotion information you need.

Moreover, based on the knowledge graph constructed by products, it can complete the question and answer more accurately than the common FAQ system, and realize the capabilities of product comparison, product FAQ support, attribute query and so on.

So how does the knowledge graph liberate human power and make intelligent customer service so good?

The mainstream knowledge graph construction method in the industry is based on the internal data and public data of enterprises, and the map service providers help customers customize the knowledge graph in the form of solutions. This method has high cost, low efficiency and long production cycle. When an e-commerce enterprise completes the knowledge map of a product, the main product may become outdated and unsellable.

In order to provide pipelin-like map construction capability, Huawei Cloud Knowledge Graph cloud service abstracts map construction into ontology construction, data source configuration, information extraction, knowledge mapping, and knowledge fusion.

Because each process module is abstracted into a plug-in form and the map construction task is generated through the combined configuration, only modification of the plug-in configuration is required to complete the construction of the enterprise knowledge map of different domains. Meanwhile, based on pipeline design, the knowledge graph cloud service can complete the update operation on the premise of only modifying the data source, which is very suitable for the knowledge graph that needs frequent update. As for how to construct ontology, configure data sources, complete information extraction and knowledge fusion, this paper is limited in space. For details, please refer to “Frontier Technology Exploration: Knowledge Graph Construction Process and Method”.

To sum up, from graph database to knowledge graph to intelligent customer service, the products and services recommended by e-commerce platforms are more and more appealing to your appetite every year on June 18. Intelligent customer service can understand you in a second after only two or three conversations, which is also the reason why consumers are immersed in the shopping carnival.

With millions of people ordering at the same time, why is it getting easier to grab the second kill?

When consumers are captured by e-commerce recommendation systems, how do they ensure that you can buy your favorite goods anytime and anywhere during the promotion period, and how do the transaction data flow in an orderly way to ensure that you can both grab and receive the goods in time? This article breaks it down for you.

Click to follow, the first time to learn about Huawei cloud fresh technology ~