In order to speed up general artificial intelligence, ant graph calculation broke the world record of authoritative evaluation for four times in a row.
Recently, the International Association data benchmark Commission (Linked Data Benchmark Council, hereinafter referred to as LDBC) released the latest results of the map data benchmark evaluation "LDBC SNB-BI". TuGraph Analytics, a flow graph computing engine developed by Ant Group, successfully completed the benchmark test on a 30TB-scale dataset, and the data scale and performance broke the previous public record of a chart database manufacturer in the United States. the concurrent throughput in the key indicators was increased to 2.84 times, and the query capacity was increased to 1.86 times.
(photo: LDBC officially released a new record of ant LDBC SNB-BI evaluation)
In this evaluation, the test product needs to quickly import and analyze data on the scale of 30TB, process data with a scale of 72 billion points and 5400 billion edges, and return the read query results within 60 seconds. TuGraph Analytics has successfully passed these stringent tests and verified the world's leading data processing capability of domestic map database products.
Graph intelligence technology, represented by graph database, graph computing and graph learning, is a new technology in the field of data and artificial intelligence, especially good at dealing with the association of massive data, excellent speed and accuracy. Gartner, an authoritative consultancy, predicts that mapping technology will be applied to 80 per cent of data and analytical innovation by 2025. In the tide of AI large models, graph intelligence is considered to be one of the key technologies to solve the illusion of large models and improve security and reliability.
According to reports, Ant began to lay out map technology in 2015 and created a leading graph intelligence technology system, including graph database, graph calculation, graph learning, knowledge graph, etc., broke the world record of LDBC map database evaluation four times, and won the first prize of Wu Wenjun's artificial intelligence, the first prize of scientific and technological progress of China Electronics Society, and the "leading scientific and technological achievements" of the World Internet Conference. TuGraph Analytic is an important component of the ant large-scale graph computing system. The second delay calculation on the "graph" of hundreds of billions of data has successfully solved the industry problems such as difficulty in financial scene risk analysis, low recognition rate, poor timeliness and so on. TuGraph-Analytics was officially open source in June 2023.
The International Association data benchmark Commission (LDBC) is an internationally recognized benchmark guide maker and test standard publisher for graph data. SNB (Social Network Benchmark) is an authoritative benchmark evaluation based on social network data, which mainly includes two sub-tests: SNB-BI and SNB-Interactive. Among them, SNB-BI mainly evaluates the ability of graph database to handle complex analytical tasks, while SNB-Interactive focuses on the ability of highly concurrent transactional tasks.
