 In this video, I will talk about our work on Interactive and Personalized Search and Recommendation for Health Empowerment. This is joint work being done within the RPI-IVM Heal Center, which involves several faculty from RPI and students from RPI and several researchers from IVM. First, a few words about the Heal Center. So Heal stands for Health Empowerment by Analytics, Learning, and Semantics. Our main idea here is to use machine learning and semantic techniques, knowledge graph methods, mining methods to basically enable health empowerment. Our main goal is to enable personalization via machine learning and semantic methods, and we're going to do this via what we call behavioral phenotype mining. So we're going to learn from users' behaviors to enable better search and recommendation. For example, for a particular disease like prediabetes or let's say prehypertension, we want to give them better search results based on their personal context, better recommendations for exercises and diet, and so on. Another goal of this project is to enable behavior change for the long term, so we want to be able to give the user tools to what we call self-experiment, that is to design actionable dietary plans, exercise plans so that they can maintain health over the long run. This slide shows the overall conceptual architecture of our system. We have multiple sources coming in, so from the personal health devices, FitBits, mobile devices, we'll collect users' personal data. Using online authoritative websites like WebMD, American Diabetes Association, we'll collect lots of textual sources to create what we call the disease knowledge graph. From the personal data, we'll create the personal knowledge graphs, and we'll also add in knowledge graphs for diets, for exercises, and so on. On the left-hand side, you see the main user interaction system where the user will converse with an app and ask questions and look for recommendations, and all this will be powered by machine learning and also technologies for conversation agents, natural language understanding, for which we're using the IBM Watson services, and we'll enhance them with newer techniques as we develop them. Here's a quick demo that showcases some of the capabilities our team has built on to the chatbot. First, we will converse to the chatbot from the perspective of an AIDS-in-url female. Now, if we go into the app and we're going to ask the chatbot, our app was good for me. The chatbot will respond by answering your question and providing additional information such as grows your age, eat one and a half cups of fruits, and two and a half cups of vegetables daily. Now, if we were to move on to another perspective, this 57-year-old male, and we were going to ask the chatbot the same as that question, our app was good for me, the chatbot will reply with another response. So now, let's say the app was good for you, but also menu age should eat two cups of fruits and two and a half cups of vegetables per day. So as you can see, the response has changed based on the age and gender of the user. As one initial step in the technology development process, the Heels Project One team conducted a series of design thinking workshops. Design thinking takes a systematic human-centered approach to problem solving. The primary goals of our design thinking workshops were to generate ideas about target user's needs and potential solutions as well as to prioritize technology goals. We conducted workshops in person at RPI and IBM Research and also online using Mural. The main outcome of the workshops was the creation of three technology goals, which focused on activity scheduling, debating information, and social community engagement. So to conclude, we have already built a prototype system where the user can converse with the app and ask different kinds of questions as we show it in the demo. And then based on the design thinking workshops, we have identified two key aims for our project. So we want to build capabilities for debating information, basically asking the system whether or not a certain activity or food is good for you, and also a system for recommending activities, exercises that people can do within, you know, based on the calendar and the personal preferences. So we'll be building a complete system that basically builds all kinds of knowledge graphs and then uses machine learning and mining combined with the semantic methods to enable an effective system for personalized search and recommendation for health empowerment.