SIRUP: Search-based Interactive Recommendations with User Profiles


Our book recommender system offers personalized suggestions based on the concise textual profiles assembled from the users' book reviews.
The users can explore different ways to create their crisp profiles from a vast amounts of unstructured input (e.g by selecting most salient keyphrases).
As a seed for recommendation, you can ask for suggestions from a preferred genre, or similar to a particular book, or matching the free-text query.

User configuration

Select a predifined user        

Please choose one of the below procedures to create a custom user profile.


Write your profile in the textarea below (text will be truncated at 128 tokens):

Select profile construction method  

Enter your book review history and choose one of the profile creation methods to create the profile:

Book Title Review
 
You can also upload your review history from .csv file and save it (2-column csv, for book title and review)



All user reviews (train set)

Define recommendation context

How to make a prediction        

write a query

select an item
select a genre
select a model (better to be the same as profile construction method)

Your profile