MSME can offer user-centric hyper-personalized recommendation service
SEOUL, South Korea, July 11, 2022 /PRNewswire/ — According to Korea IT Times, Harex InfoTech, Inc. announced on the 10th that they attended the “2022 Spring Conference of Korea Management Information Society ” held at Hanwha Resort Haeundae Tivoli in Busan from June 9 to 11 and announce the results of a “new B2B service using a user-centrichyper-personalized recommendation system” conducted with a research team led by Professor Kyoung Jun Lee from Kyung Hee University (Department of Big Data Application).
Professor Kyoung Jun Lee’s research team (Professor YuJeongHwangbo, researcher Baek Jeong, Soohyun Kim, and Gunho Lee) and Harex InfoTech’s User-CentricArtificial Intelligence Research Institute (Manager Youngjae Cho, Vice President Moonho Yang) proposed new B2B target marketing service and new product brainstorming using their own developed user-centric hyper-personalized recommendation system. The recommendation system is usually used in a B2C service that recommends products to users, but Harex InfoTech’s recommendation system is a B2B service methodology that can be used in stores.
The research team utilizes a list of recommended products derived as a user-centrichyper-personalized recommendation system for the target marketing. The methodology is to make user lists by products after cross-linkingthe lists of the recommended products for each user by product, and finally, target marketing the listed users by products. This technology is currently under patent-pending.
For example, if ‘Company B’s Fried Chicken’ is derived as a recommended value based on user A’s purchase history, then company B can conducttarget marketinga fried chicken on user A. Existing target marketing was mainly conducted in groups based on demographic information such as women in their 20s and men in their 50s, but the research team’s recommendation system allows hyper-personalized target marketing based on duality (pair or confrontational relationships).
For brainstorming of new product, the research team uses a user-centrichyper-personalized recommendation system that utilizes the product name in natural language. Since the AI learns by morpheme unit (separating ‘Ham cheese toast’ into ‘ham/cheese/toast’), which is the minimum unit of the product’s name meaning, it is possible to derive a product name that does not actually exist. Derived products that do not exist in reality may rather be more suitable for users. It is under patent to use this idea for the development of new products.
It is unique that this methodology uses morpheme units to study. For example, if a user purchased ‘Chicken Breast Cream Spaghetti’, ‘Octopus Bibimbap’, ‘Stir-fried Spicy Pork’, and ‘Ham Cheese Toast’, they would learn it in the form of ‘Chicken Breast’, ‘Cream’, ‘Spaghetti’, ‘Octopus’, ‘Bibimbap’, ‘Stir-fried Pork’, ‘Rice’, ‘Ham Cheese’, and ‘Toast’. The derived value is also combined in morpheme units. Non-existing product names such as “Octopus Cream Spaghetti” and “Spicy Pork Toast” can be derived. The new product name derived in this way can be presented as a brainstorming idea for the development of new products.
Previous recommendation system research was used as a single service task called recommendation model, but Professor Lee’s research team, through one recommendation system, utilize it as a multi-service task such as B2B service-based target marketing and new product development brainstorming as well as hyper-personalized B2C recommendation service.
Researcher Baek Jung said, “The large language model GPT-3 released by OpenAI has become the one service engine to utilize various application services such as Q&A, grammar correction, and chatbot, etc.” and “We will develop various service tasks by utilizing the user-centric hyper-personalized recommendation system that has been developed as an AI engine and makeit to be used in the general commercial environment.”
Professor Kyoung Jun Lee, who leads Harex Infotech’s User-centricArtificial Intelligence Research Institute, presented under the theme of “Artificial Intelligence Sharing Platform as an Alternative to Exclusive Platforms” at the 2022 Spring Conference of Korea Management Information Systems. An artificial intelligence sharing platform based on User-CentricArtificial Intelligence is one of the alternatives against exclusive platforms from big tech or large companies.
Companies such as Big Tech monopolize data secured by customers, small business owners, individual businesses, and small and medium-sized enterprises. Professor Lee uses newly developed federated learning technology to pursue synergy in which customers and small and medium-sized businesses share artificial intelligence while keeping their data intact. Based on shared artificial intelligence, it presents a federated platform business model that provides and shares various services such as recommendation, target marketing, new product planning, and payment services.
Conventional federated learning is a form of sharing AI by creating a single model through individual companies’ data, but the newly proposed federated learning method divides data from individual companies and creates a final AI sharing model. The new federated learning methodology has been patented
Professor Kyoung Jun Lee said, “The largest AI engine at each industry and level, including medical, commerce, transportation, finance, smart farms, manufacturing, robots, and smart cities, will be built by AI sharing methods,” and “If we introduce this method, we can use each other’s data without signing a data agreement between economic players, so we will make it possible of new cooperation and business models that were previously impossible between individuals, companies, and institutions.”
Kyung Yang Park, Founder, President and Chief Vision Officer of Harex InfoTech Inc. said, “In the beginning, the business will start with a small size of AI sharing between several companies, but it will expand gradually,” adding, “It will be an opportunity to change the business model and collaboration method completely, and furthermore, this system will be established as a sustainable way of business in the era of stakeholder capitalism.”
Harex InfoTech announced the world’s first mobile card payment service in Silicon Valley in 2000 and 2013 won the ‘Best Technology Award’ at the Innovation Project 2013 held by PYMNTS.com at Annenberg Hall on the Harvard University Campus, and is building the User-Centric Hyper-Connected Shared Network by sharing the platform with all businesses and based on the network created, is creating a user-centric new economic order that maximizes user benefits and business profits with hyper-personalized Digital Me services based on User-Centric AI sharing platform.