Automatic recipe creator (MSc/BSc)
Open datasets such as Open Food Facts, are a great source of information. Within FrieslandCampina we are constantly challenged to use information such as this to innovate on our business process. This project will deal with developing a proof of concept of how the open food facts system can be used to generate a recipe suggestion tool. The tool should use the input of preferred ingredients and based on this display related products from the database and suggest a possible next ingredient to be added and its possible concentration.
During a product development process, a developer has to take many choices below which the selection of the right ingredients for a specific product is highly relevant. With open sources such as open food facts available, it would be great to have this knowledge at our fingertips during this step.
This project at FrieslandCampina therefore intends to develop an algorithm based on the open food facts dataset where a developer can start with one ingredient and a target product category and the algorithm suggests a logical next ingredient (and possibly the weight percent thereof). The topic is challenging not just because of the data cleaning and preparation involved but also because of the variety of approaches that are available to solve this task from which the successful candidate needs to choose the most relevant one to achieve the expected goal.
- Review previous work on related approaches using open data ideally in the particular area of recipe suggestions
- Design and implement a proof-of-concept tool that uses Open Food Facts dataset to suggest recipes based on given preferred ingredients
- Evaluate the effectiveness of the solution using a real case study at FrieslandCampina
The work in this thesis entails:
- Scan the internet and relevant scientific literature for possible approaches to create a recommendation algorithm and select the most promising approach
- Prepare open food facts data and feed to the recommendation algorithm (training)
- Develop a simple front-end that can be used by a product developer as a proof of concept.
- Open Food Facts food products database https://world.openfoodfacts.org/
- Courses: -
- Required skills/knowledge: Data ingestion / wrangling / cleaning, algorithm development (i.e. machine learning), simple front-end development (e.g. Shiny / Bokeh, etc.) are nice to have, otherwise to be acquired during the thesis
Key words: Open data, recommendation algorithms, machine learning
- Önder Babur (email@example.com)
- Carsten Ersch (Carsten.Ersch@frieslandcampina.com)