Customers need personalized meal suggestions that are both in line with their health goals and their food tastes. No matter if the customer is on low sodium, low-cholesterol or on a fully personalized DNA profile-driven diet they can get the right suggestions. Is your customer in need of only recipes that are vegetarian, contain less than 600mg of sodium per serving, are high on Vitamin A but low in fiber? PHILIA RECIPE ENGINE can help with this.

What it does

PHILIA Recipe Engine was originally an Alexa conversation skill, which provides more than 2 million indexed, standardized, and complete nutritional information recipes; so I decided to create a facebook-messenger bot for this project as well. The bot allows searches by amount of nutrients, 40 diets and health labels, as well as keyword searches by type of cuisine (Chinese, Italian, Indian, French, ...), by type of meal (lunch, dinner, breakfast, snack) or by type of dish (soup, salad, pizza, sandwich, ...).

How I built it

It is not a basic or easy bot to implement, as it is developed with innovative technologies such as:

  •, Facebook's open source product for building Natural Language Experiences.
  • Facebook-Messenger, Facebook Messenger is an instant messaging system created by the company Facebook, and incorporated into the Facebook social network. -Philia Recipe Engine API, Recipe Search API, Food Database API, and Nutrition Analysis API
  • Node js , a JavaScript runtime built on Chrome's V8 JavaScript engine.
  • Heroku, a platform as a service (PaaS) that enables developers to build, run, and operate applications entirely in the cloud.

Challenges I ran into

Each task / MVP brings new challenges, testing the effectiveness of individuals or a team. It takes a lot of effort and work to complete an MVP. In addition to technical and developmental skills, completing a task also requires many other skills. It tests an individual's abilities, team coordination, and communication skills.

It is a very difficult task to recommend food to people mainly based on their nutrient needs, because people like different types of dishes. It varies depending on where they come from, where they live now. In addition, a person's budget is very important for the type of food they prefer over nutrition. But we have optimized our result to obtain very precise recommendations.

Accomplishments that I'm proud of

We are proud to be able to say that we finally managed to accomplish our task with decent results, as the saying goes "hard work pays off". It's the complete technology for all the time that we update. Its full MVP, many more features will come soon. We continue to update it with better results.

What I learned

I learned many skills such as :-

  • How to avoid overequipment in machine learning models.
  • how to break down a complex task into several parts and steps
  • Refine understanding of challenges through discussion and explanation
  • How to apply development skills in the real world for the benefit of the world

What's next for PHILIA Recipe Engine

We plan to improve its accuracy to make it more user-friendly by following these steps:-

  • Trying to merge the most appropriate recommendations according to the origin and quantity of the recommended diet to obtain optimized results that appeal to the client.

  • Increasing the variety of dishes and user-specific recommendations.

  • and add a voice interface to our bot

Currently we cover the food data of some regions in a recommendation; so we cover a small region (USA, UK, Italy, France, Spain, China, India, ...) but soon we try to increase a wide variety of foods in our bot especially African meals.

Share this project: