2021WAIC | Individual Push CTO Ye Xinjiang: Data Intelligence under the Trillion-level Graph


Recently, Ye Xinjiang, CTO of Daily Interaction (a tweet), was invited to attend the WAIC World Artificial Intelligence Conference, and delivered a speech at the "Graph Database Technology and Application under Big Data Association" theme forum, discussing "Trillion Billion" with participating experts and audiences. Data Intelligence under the Level Diagram".

Ye Xinjiang introduced that Daily Interactive was established in 2010 to provide APP message push services. For more than ten years, I have participated in and witnessed the rapid development of the mobile Internet industry every day. Relying on its own massive data resources and the advantages of big data and artificial intelligence technology, it has built a complete data intelligence service ecosystem and provided professional big data solutions. , To promote the digital and intelligent upgrading of the mobile Internet, brand marketing, finance, smart city and other sub-fields.

At present, Daily Interaction has grown into a new economic complex that organically integrates the characteristics of emerging fields such as the Internet, big data, artificial intelligence, and cloud computing, and strives to create a data center product-the "daily rule of number platform". Output the ability to regulate numbers to increase energy and efficiency for digital innovation in vertical industries.

The new development of data intelligence: solving the problem of big uncertainty in reality

Daily Interaction takes "data to make the industry smarter" as its mission and vision, and has its own unique understanding of data intelligence. Ye Xinjiang said that in the information age, we mainly use data to describe objective reality. For example, we use a large visual monitoring screen to describe the road conditions, and use different colors to represent the degree of congestion on the road. Later, we increasingly used data for diagnosis and causal analysis, such as attributing road congestion. In recent years, data has continued to explode, and cutting-edge technologies such as machine learning and graph mining have been widely used, and data intelligence has developed to a new stage. At present, people's application of data is not only at the stage of description and diagnosis. People hope to use data intelligence to solve the problem of large uncertainty in reality, and to predict the future, so as to grasp the situation, grasp the trend, and grasp the initiative.

The correct way to solve problems: ontology modeling and retrieval

Ye Xinjiang mentioned that problems with high uncertainty in reality are often problems in an open environment and are affected by many factors. In order to solve this kind of problem, the traditional deep learning method based on neural network requires a large number of parameters to model the environment. For example, the recently popular GPT-3 model contains hundreds of billions of parameters, and the cost of one training reaches tens of millions. US dollars. Even so, it is difficult for artificial intelligence in this form to reach the level of human intelligence.

Therefore, we judge that the final form of artificial intelligence should be a "human brain + computer" man-machine symbiosis method. How to achieve "human-machine symbiosis" to solve these uncertain problems? Data intelligence based on the knowledge graph is a promising direction. On the one hand, the existing knowledge is digitized through ontology modeling, so that the computer has the way of thinking of the human brain; on the other hand, through retrieval and reasoning on the knowledge graph , The human brain can use the computing power of the computer. In order to achieve this goal, the underlying infrastructure needs to meet the requirements of the modeling architecture, and have the capabilities of fast retrieval and global derivation. The comprehensive graph database system that integrates graph query system and graph computing system can meet these characteristics and functional requirements.

Data intelligence practice under the trillion-level map: big data fights the epidemic

So, how to carry out data intelligent application based on the comprehensive graph database system to solve the problem of large uncertainty in reality? Ye Xinjiang used daily interaction to participate in the big data anti-epidemic as an example, and shared the data intelligence practice of daily interaction under the trillion-level graph.

After the outbreak of the new crown pneumonia in 2020, the big data anti-epidemic team-"personal doctors" was established interactively every day, and cooperated with the team of Academician Li Lanjuan to participate in this battle against the new coronavirus, and to study and judge the epidemic situation. Conduct in-depth research on spread path analysis and other aspects to comprehensively assist the precise prevention and control of the epidemic.

In order to help local governments achieve efficient epidemic prevention, the company and the team of academician Li Lanjuan proposed the concept of "unconscious close contacts", based on human space-time big data, to help relevant departments find key areas, key populations and key scenes to achieve intelligent prevention and control. At the same time, in order to facilitate local governments to comprehensively understand the epidemic situation, we use big data to reflect and quantify the current epidemic risk in the region, and provide strong data support for efficient epidemic prevention and control. In order to help local governments to promote the resumption of work and production in an orderly manner, we have also participated in the development of the health code coding engine, which integrates the information of the three dimensions of "space, time, and the human world" to calculate the close connection risk, combined with the current prevention and control strategies, Help complete the final code.

In fact, the above applications all rely on the comprehensive graph database to efficiently model, retrieve and reason about the relationship of the crowd in the three dimensions of “space, time, and humanity”, and the superposition of the three dimensions forms the final ten thousand The billion-level map expands the application scenarios of data intelligence in social governance and smart medical care.

to sum up

Today, economic development presents a new paradigm, and data has become a new type of production factor, which will play an important role in driving future development. As an important infrastructure in the era of data intelligence, graph databases have created sufficient conditions for us to perform complex calculations such as dynamic retrieval, statistics, and relationship derivation in multiple dimensions such as time and space in the entire domain.

In the future, daily interaction will continue to practice based on technologies such as graph databases and knowledge graphs. By tapping the potential of data, releasing the power of data interconnection, promoting the solution of uncertain problems in reality, and making greater contributions to industrial development and social progress the power of.

阅读 706



1.5k 声望
2.4k 粉丝
0 条评论


1.5k 声望
2.4k 粉丝