Prof. Wong Kam Fai Develops AI Integrated Community-based Dining Recommendation System

Media Release
Hong Kong has a reputation as a gourmet paradise and there have been numerous dining guides and review platforms offering a wide range of recommendations for tourists.  A team led by Prof. Wong Kam Fai, Department of Systems Engineering and Engineering Management has recently developed a personalised recommendation system INCOMIRS (Integrated Community-based Microblog Recommendation System) based on historical posts and reviews on social networks and established digital dining guides. The team has spent two years making this recommendation system a useful and multilingual tool for anticipating users’ preferences in restaurant locations and dish types.  Its Android app will soon be made available for free download, offering tourists an intelligent way to explore good food in Hong Kong.  The project has recently received the first-class award in the innovation category of the 3rd Chinese Youth Congress on Artificial Intelligence. 

Currently, most popular dining guides generally make recommendations according to the ratings by public users, thus the restaurants are usually famous ones or liked by most people.  The peer impact is not considered as an indicator for making a specific recommendation in existing dining guides.  However, people in reality tend to pick restaurants that are posted on social networks and recommended by peers.  Also, traditional dining review algorithms focus more on the individual consumer behaviours than on the community consuming pattern.  It is therefore a breakthrough for Prof. Wong’s team to better utilise social network resources to assist users in decision making.  By integrating data mining, translating function, and software application, they have designed this novel algorithm to collect and propose a more personalised search result and preferred dining recommendations for each different user, based on their social networks.  The results can also minimise the advertising impact or the information overload phenomenon. 

The recommendation system is easy to use.  It will run automatically once the app is downloaded with a linked account to social networks.  In principle, restaurants with more interactions and reviews on social networks will be prioritised higher than others. If users do not log into the system, they can still simply browse the popular restaurant chart and food trends for preliminary recommendations. 

In addition to the English and Chinese versions, the recommendation system is tailored to fit different language needs of the world’s travelers, specifically Japanese, Korean, French, Russian, and Spanish.  Foreigners can make use of the dining app to find good local food.  According to the 2017 visitor arrival statistics from the Hong Kong Tourism Board, there were over 1.23 million Japanese and 148,000 Russians visiting Hong Kong, which was a 12.6% and 3.8% rise, respectively compared to 2016, while Spanish is regarded as one of the official languages of the United Nations and has over 500 million speakers worldwide. The unique and multilingual function of INCOMIRS will help promote Hong Kong as a gourmet paradise with a great variety of creative culinary delights. 

People today are generally very fond of travelling. The research team is planning to develop another useful app for tourists to plan travel itineraries and generate more recommendations from other data which is also available on users’ social networks. For example, the numbers of photos on dining, shopping and sightseeing will be adopted in the system. ‘The recommendation system is a part of the CUHK’s smart innovation and technology products. We strive to transfer our knowledge and promote information technology research and application for the benefit of mankind,’ said Prof. Wong Kam Fai. 


Prof. WONG Kam-fai, Associate Dean (External Affairs) of the Faculty of Engineering and Professor of the SEEM Department (middle); Dr. Gabriel FUNG Pui-cheong (left) and Mr. LI Bo from the SEEM Department.

The "EatChoice" App