Skip to content
/ ARMED Public

Analogy-based Recommender system of Multi-Ethnic Dishes

License

Notifications You must be signed in to change notification settings

uazadi/ARMED

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

18 Commits
 
 
 
 
 
 
 
 

Repository files navigation

ARMED

Analogy-based Recommender system of multi-ethnic dishes

Recently, even in the smallest town of Italy it can be found an increasing number of restaurants that offer dishes originated from all over the world. With such a heterogeneous offer it is likely that the average person will not know a priori which type of cuisine or more specifically which dishes will be likely to meet his/her taste.

The purpose of this project is to support the customers in this specific decision making step, by suggesting them a dish of a cuisine never tasted before based on their preferences in a known cuisine, as for example the cuisine of their home country.

The main features are:

  • Exploit a multi-modal embedding representation of dishes in order to accomplish the analogical inference through which the recommendation is obtained;
  • Keep into consideration specific diets such as: vegetarian, vegan, dairy-free, ...
  • Allow the users to add their own recipes and use them to obtain the recommendations

What is shared through this repository?

  • The scripts that accomplish the crawling process through which the recipes have been downloaded from BBC Food (available here);
  • The jupyter notebook through which the recepies and images embedding can be explored (available here);
  • The "ready-to-use" application comprehensive of the GUI that allows to exploit the main functionalities described above (available here)
  • Furthermore all the data can be downloaded from here (link to Google Drive)

Further Information

If you want to know how this work, it is available the presentation.

Furthermore, you can contact me at the following email address: u.azadi@outlook.com (Please insert ARMED in the subject)