The new generation of information technology led by AI is driving a new wave of science and technology. As one of the fastest-growing mobile Internet platforms in China in recent years, Xiaohongshu has taken advantage of the momentum and has now formed a very large UGC community focusing on graphic, text and short video content. In this unique and active community, massive multi-modal data and user behavior feedback are generated every day, giving rise to new problems that are both valuable and challenging.
Currently, many exciting developments are taking place in large-scale deep learning systems. At the "Xiaohongshu REDtech Youth Technology Salon" event on October 15, 2022, Cage, vice president of technology at Xiaohongshu, shared "Large-scale deep learning system technology and its application in Xiaohongshu" and revealed to us The "mystery" of LarC.
Cage: Vice President of Technology at Xiaohongshu. He graduated from Shanghai Jiao Tong University. He once served as Vice President of Technology at Huanju Times and Chief Architect of Baidu Fengchao. He was responsible for Baidu search advertising CTR machine learning algorithm work. He once served as the China technical leader of IBM's Deep Question Answering (DeepQA) project.
The following content is compiled based on Cage’s on-site report
Xiaohongshu Business Overview
Sharing real life experiences of ordinary people
Xiaohongshu is a booming content community where a large number of people who understand life and love to share exchange their life experiences and attitudes, and it continues to attract more and more users to join. Now, Xiaohongshu has 200 million monthly active users, of which more than 70% are born in the 1990s. 50% of users come from first- and second-tier cities, and half come from third- and fourth-tier cities. The composition of users is very rich and young. .
"Ordinary people" are sharing their "real" "life experiences", which is a very big difference between Xiaohongshu and other content platforms and communities. First of all, the sharers are "ordinary people". Secondly, "sincere sharing and friendly interaction" are the Xiaohongshu community conventions, and "sincerity" is a very important point. Sharing in these communities is closely related to our offline life consumption, such as Treasure Bookstore, or how to dress, decorate, cook, etc., which are everyone's daily "life experience".
We can also use some numbers to measure the development of the Xiaohongshu community over the years. We see that the number of notes published has been growing at a very fast rate every year from 2018 to 2021. From 2020 to 2021, Xiaohongshu users The number of notes published increased by more than 150% year-on-year.
Three main businesses: community, commercialization, and e-commerce
In such a rapidly developing content community, the three main businesses are community, commercialization and e-commerce.
First of all, our content community and content platform is a lifestyle content community covering all life categories and focusing on UGC . Also because of this kind of "sincere sharing" that fits life and daily consumption, users have a high degree of trust in our community content. Everyone will be "seeded" when they see good lifestyles, consumer content, services and products, etc. "Grass", we bring brand and effect transformation through our unique "grass planting" business model .
"After planting grass, can we pull it out?" While consuming content, everyone also hopes to be able to buy their favorite items naturally and conveniently. This is our efficient closed-loop consumption field , which is the e-commerce part. .
Xiaohongshu Technical Challenge
Multimodal technology is one of the technology directions that has attracted widespread attention and is developing rapidly in the entire AI field. The UGC community and content ecology contain a large amount of images, videos, text and user behavior information, resulting in a massive amount of high-quality multimodal information. Data therefore becomes an excellent practical scenario. Users like good content when they see it, perform various search behaviors, watch a certain video, etc., which constitute a large amount of actual user feedback.
Nowadays, the number of feedback samples actually generated through user behavior every day is in the tens of billions. How to mine content that users are interested in and good commercial content in massive multi-modal data . Starting from this goal, many valuable and challenging problems have been derived.
How we solve these technologies:
Real-time recommendation system for thousands of people
When you open Xiaohongshu, the first thing you see is the waterfall flow or content flow listed. These are the contents recommended by the recommendation system. According to statistics, Xiaohongshu generates tens of billions of user actions every day. For this data, the Xiaohongshu technical team uses a machine learning framework based on LarC to train the model, and based on the rules in user behavior, it finds the content that the user is interested in and recommends it to the user.
The figure below shows the general structure of the Xiaohongshu recommendation model. This is a multi-task machine learning model that can predict the user's clicks, dwell time, whether to like and collect, etc. In view of the massive coefficient parameters generated by the Xiaohongshu platform, Xiaohongshu updates and captures these parameters through a very large-scale conflict-free parameter server.
Online training of the recommended system is as follows. When users browse the information flow, the recommendation system will capture the user's browsing, clicks, likes and other behaviors in real time. These behaviors will be spliced based on Flink's real-time processing computing engine to generate high-performance samples. Samples will be sent to the model in real time for prediction. At the same time, these short-lived accumulated samples will also be used for a very short online training to update model parameters. These updated model parameters will be published online immediately to serve the next request. The entire process is kept within minutes.
There is also a classic question in the industry. For example, when people browse recommended content, they often find: Why are things that I have seen before intensively pushed? What should I do if the things I watch are not fresh enough?
In recommendation scenarios, focusing on shorter time periods will lead to serious problems of chasing and information cocooning. Xiaohongshu’s technical team has designed different sequence modeling methods for users’ diversified long- and short-term behaviors, bringing benefits in multiple dimensions. significantly improved. In addition, regarding the diversity issue of content recommendation, Xiaohongshu’s technical team improved the traditional diversity approach from DPP to SSD algorithm, and efficiently calculated the sliding window in the information flow recommendation scenario, thus transforming the value ranking of single article models. Model the entire browsing cycle. What this relies on is the twin neural network learning the similarity of long-tail content.
We have published related work results at the KDD 2021 conference. It has transformed from the estimation of the value of a single article to the estimation of the value of a sequence, and from the diversity of a single article to the diversity of multiple articles. It is also based on the SSD algorithm, and Evaluation of content similarity based on this Siamese neural network.
Multimodal generalized life search engine
Because the Xiaohongshu community contains a large amount of very useful information in real life, many users use Xiaohongshu as a search engine. This includes some challenges, such as searching in multiple data forms, serious long-tail phenomena, and intent understanding issues.
Existing image and text search engines can search for pictures through text, but the method is relatively simple. Usually, the pictures are tagged with text, and then the text is matched. The next-generation multi-modal pan-life search engine built by the Xiaohongshu team is based on an in-depth understanding of multi-modal content. It can truly search for visual content through images, text and text, and can also make more personalized searches based on the characteristics of users. search.
What is a pan-life knowledge search engine? For example, we see a good-looking piece of clothing or shoes on Xiaohongshu and want to search for its combinations and how it looks in different situations. This is a search for life knowledge, and it is also a multi-modal search.
What this shows is the multi-modality planned by the Xiaohongshu technical team, especially for technical architecture such as image search. One of the key dependencies is the feature multi-module, which requires large-scale neural networks for representation learning. The content contained in the picture, whether it is clothes, shoes or other merchandise, can have a good representation. It is very good to retrieve the same products or similar products from a large amount of multi-modal content. This is an application of our large-scale neural network in search.
AI generates more native business content
Compared with other platforms, Xiaohongshu’s commercial content has one big difference – nativeization. The so-called nativeization means that from the perspective of likes, comments and other behaviors, users appreciate the content very much and may not feel that it is commercial content at all. But for merchants on the platform, the threshold for producing such commercial content is very high. How to strike a good balance between the business intentions of merchants and the user value of content produced is a critical issue.
To this end, the Xiaohongshu technical team uses generative technology based on large-scale neural networks to help merchants generate better titles and content based on the content. For example, merchants can choose to express multiple selling points, or they can choose to highlight target customer groups, or their favorite Xiaohongshu style. The machine will automatically give suggested titles. After quoting the titles created by the machine, regardless of business effects, clicks or The length of stay has been greatly improved, and users also like this kind of content very much, so it achieves a good balance between business and user value.
Behind this is actually based on large-scale pre-training models, including the industry's leading model architectures such as T5, BERT, and GPT. These model architectures are trained on Xiaohongshu's massive multi-modal data. Part of the pre-trained model is used to understand the content of notes, and part of the pre-trained model is used to guide the generative model to generate titles. These are how related technologies are applied in the business field.
Large-scale machine learning platform
All the above machine learning content is actually based on the LarC machine learning platform self-developed by the Xiaohongshu technical team. It was launched in 2019. By 2020 and 2021, related machine learning frameworks and platforms were promoted to all fields such as search, recommendation, and advertising. In 2022, LarC will become a platform.
At present, the capabilities of the LarC machine learning platform are quite complete, covering multiple levels from underlying infrastructure to computing framework, resource scheduling, offline applications, and online deployment (the yellow part represents that it has been implemented).
With the help of the LarC machine learning platform, the Xiaohongshu technical team hopes to help all algorithm students quickly and efficiently process massive data and train large-scale machine learning and deep learning models.
Summary
Xiaohongshu is a rapidly developing content community. “Ordinary people”, “real sharing” and “life experience” are its keywords.
In such a scenario with massive multi-modal data and user feedback data, many cutting-edge technology explorations have been spawned. The above is a selection of some points from a large amount of technical work to share with you. In fact, there is a lot more content. I hope everyone can understand Xiaohongshu's technology and large-scale deep learning from it.
"Q&A" session
Q: The current diffusion model generation capability is very strong and attracts a lot of attention. Does Xiaohongshu explore the application of this technology?
Cage: At present, in our commercial content creation process, we have used generative models to help merchants generate more native content and generate content that is more in line with the tone of Xiaohongshu. If you truly understand Xiaohongshu’s business , it is well combined with the model, and it is actually very popular among users, so I think this type of generative model will definitely be widely used in Xiaohongshu in the future.
Q: How do young technicians maintain their technical competitiveness after employment? What is Xiaohongshu’s training plan?
Cage: For outstanding fresh graduates, in addition to providing rich resources such as scenarios and data, the Xiaohongshu technical team has formulated a detailed training plan for the entire cycle from integrating into the workplace to growing into industry technical talents to escort every The growth of top AI technical talents in universities.
In the first year, the Xiaohongshu technical team focused on "integration" and helped everyone complete the transformation from students to professionals through the Shuguang Plan, Mentor mechanism and other methods, while helping everyone find the technical direction they are good at in practice.
In the second and third years, the Xiaohongshu technical team will train young talents to become unique technical backbones through business practice, systematic courses, cutting-edge sharing and academic exchanges. In this process, outstanding students also have the opportunity to grow into technical leaders.
In particular, for outstanding fresh graduates after joining the company, their immediate superiors will always pay attention to the growth process of the newcomers, provide guidance and help, and also share technical experience from major domestic and foreign manufacturers with the students.
Today, Xiaohongshu is in a period of rapid growth. The value of the technical team has been further highlighted. The demand for exploring cutting-edge technologies and their implementation is stronger than ever before, and we are looking forward to the addition of more top AI talents.
At the same time, I also hope that students can set challenging goals for themselves. The Xiaohongshu platform will continue to provide a practical battlefield for young AI technical talents.