TWeD Talk: MealQA: A food-oriented intelligent assistant

Printer-friendly version

TWeD Talk: MealQA: A food-oriented intelligent assistantSeptember 23, 2013
There's always something happening on Wednesday evenings in the Tetherless World!

TWeD Talk, Wednesday, September 25, 2013, 7pm ET, Winslow Building on the RPI Campus

During the summer internship at Samsung R&D Center, TWC grad student Yu Chen worked on the MealQA system, a food-oriented intelligent assistant which can answer natural language queries w.r.t. food and dishes. For example, given a natural language query such as "What's the best restaurant for bibimbap?" the system is able to provide a list of Korean restaurants that offer bibimbap (based on menus and other data), ranked according to sentiment analysis and entity extraction from available reviews. Yu's responsibility was to develop a ranking engine that provides the "most consistent" results at the front in the ranking list. The techniques include Bayesian networks and information theory. The dataset Yu trained the model on is based on Freebase, DBpedia and some proprietary datasets.