Beijing Chaoshan Food is a rich and varied dimension of cultural heritage, crucial to both individuals and social groups. To bridge the gap in the literature on the often-overlooked regional diversity in this domain, we introduce FoodieQA, a manually curated, fine-grained image-text dataset capturing the intricate features of food cultures across various regions in China. We evaluate vision-language Models (VLMs) and large language models (LLMs) on newly collected, unseen food images and corresponding questions. FoodieQA comprises three multiplechoice question-answering tasks where models need to answer questions based on multiple images, Sichuan Guangdong a single image, and text-only descriptions, Figure 1: An example of regional food differences in respectively. While LLMs excel at text-based referring to hotpot in China. The depicted soups and question answering, surpassing human accuracy, dishware visually reflect the ingredients, flavors, and the open-weights VLMs still fall short by traditions of these regions: Beijing in the north, Sichuan 41% on multi-image and 21% on single-image in the southwest, and Guangdong in the south coast. VQA tasks, although closed-weights models perform closer to human levels (within 10%).

I believe in most cases it makes sense for bot makers to build their own natural language parser, rather than using a third party API. There are good strategic and technical arguments for doing this, and I want to show how easily you can put something together. So what NLP do you need for a typical bot? Let's say you're building a service to help people find restaurants. In a previous post I mentioned that tools like wit and LUIS make intent classification and entity extraction so simple that you can build a bot like this during a hackathon.

The new Alexa feature allows customers to order from any restaurant available on the service by saying, "Alexa, order from Amazon Restaurants," and have any meal delivered to their door in under an hour or less. Amazon Prime members can ask Alexa to reorder from Amazon Restaurants simply by saying a restaurant name or a cuisine type. For example, you can say "Alexa, order sushi from Amazon Restaurants," The service will then pull your order from a specified restaurant or cuisine type and lists different meal options that are currently available for reorder. Once you select a meal it will then delivered to your default address. "Customers now have a hands-free, hassle-free way to reorder any meal from Amazon Restaurants using their voice to get dinner on the table," said Gus Lopez, general manager of Amazon Restaurants at Amazon, in a statement.