[단독]Consumer prices, now measured by AI… Investigation period is also shortened to one day

Using artificial intelligence (AI), an online price index has been developed that can grasp the trend of daily necessities price fluctuations on a daily basis. The government is piloting the online price index for masks and plans to expand it to 15 agricultural, livestock, and marine products and 20 processed foods.

On the 8th, a government official said, “In the mask out-of-stock crisis triggered by the novel coronavirus infection (Corona 19), we have come to recognize the importance of daily trends by item.” “Through the online price index using AI, the impact of the real economy caused by unexpected events You can prepare for it on a daily basis.” The online price index developed by statistical authorities such as the Ministry of Strategy and Finance, Statistics Korea, and the Bank of Korea using AI is planned to be opened to the public in 2024 after establishing a system from next year to 2023.

The online price index is a price trend that is grasped on a daily basis by collecting and analyzing price information published on Internet sites. By using web scraping technology, it is possible to collect 2.5 million online price information daily on a domestic basis. It is a method of collecting price information on the Internet, so it may differ from the actual price, but it has the advantage of being able to measure the trend of price fluctuations by day.

In particular, the government believes that online price surveys are important to determine inflation as consumer purchasing behavior is gradually shifting from offline to online. Another advantage is that products that are wider than the existing consumer price index (CPI) can be reflected in the index. Compared to the consumer price index that comes out once a month, the online price index can be measured on a daily basis. However, since it is a large amount of data (big data) collected online, there is a problem that even unrefined data can be mixed. So, from April of last year, the statistical authorities began to develop a system that could solve this problem through AI technology.

In advanced countries, there are increasing attempts to supplement the existing price trend survey through AI. The US Bureau of Labor Statistics (BLS) is reflecting big data collected and organized by AI in the price index of some electronic products such as TVs and cameras. The UK National Statistical Office (ONS) is also using AI to launch a new online price index (CLIP) focusing on food and beverages.

The government has started daily price trend surveys on 123 items from 2015, and it plans to use AI for 35 items. Already among agricultural, livestock and marine products, rice (10kg), glutinous rice (1kg), beef (domestic sirloin 100g), pork (domestic pork belly 100g), chicken (one), galchi (one), mackerel (large one) One, one small, two), abalone (100g), dried anchovy (1kg), dried squid (five), apple (3kg), tangerine (1kg), potato (1kg), garlic (1 ㎏), red pepper powder (100g), etc. 15 items are supported by AI. Among processed foods, kimchi (1kg), flour (hereinafter, based on packaging units), noodles, vermicelli, cereal, pancake powder, pasta noodles, milk, milk powder, sesame oil, cooking oil, canned fruit, chocolate, gum, sugar, soy sauce, red pepper paste , Vinegar, soy milk, mineral water, etc. for 20 items, AI creates an online price index.

A government official said, “Thanks to the use of AI, we were able to filter out unnecessary data such as dog masks, mask hangers, and medical nose masks from the big data of KF94·80 masks.” “Starting with items with high accuracy of 89~94%. In 2024, when the index development is finalized, AI will be applicable to most items.”

Reporter Kang Byung-cheol [email protected]