Skip to content


Switch branches/tags

Name already in use

A tag already exists with the provided branch name. Many Git commands accept both tag and branch names, so creating this branch may cause unexpected behavior. Are you sure you want to create this branch?

Latest commit


Git stats


Failed to load latest commit information.
Latest commit message
Commit time

Using Posters to Recommend Anime and Mangas

Featuring Blended Alternate Least Squares with Explanation (BALSE).


The classic recommendation problem is the following: given a user and the items (mangas) that they like, how can we recommend new items (mangas) that they are also likely to enjoy? Typically this is done via collaborative filtering, i.e. people with similar taste also enjoy other mangas, so we recommend these to the original user. A very common problem occurs when you have a new or obscure manga, aka the cold-start problem. There are no reviews to use for this manga, so a cooler option is to build a system that actually understands the content it recommends. We propose extracting visual information from the posters of these little-known mangas, using a deep neural net called Illustration2Vec. The theory is that users that like mangas with "girl with sword" will also like other mangas that have "girl with sword" or perhaps "girl with bow" but probably not "multiple boys in a swimming pool".

To cite the paper:

   author = {{Vie}, Jill-J{\^e}nn and {Yger}, Florian and {Lahfa}, Ryan and {Clement}, Basile and 
    {Cocchi}, K{\'e}vin and {Chalumeau}, Thomas and {Kashima}, Hisashi},
    title = "{Using Posters to Recommend Anime and Mangas in a Cold-Start Scenario}",
  journal = {ArXiv e-prints},
archivePrefix = "arXiv",
   eprint = {1709.01584},
 primaryClass = "cs.IR",
 keywords = {Computer Science - Information Retrieval, Computer Science - Learning,
    Statistics - Machine Learning, recommender system, cold-start, collaborative filtering, LASSO, tag prediction},
     year = 2017,
    month = sep,
   adsurl = {},
  url = {},
  adsnote = {Provided by the SAO/NASA Astrophysics Data System}

Mangaki Data Challenge

We organized a data challenge about recommender systems with Kyoto University. See the results.


If you're interested in the dataset, please contact us.

In the meantime, you can check:


Using Posters to Recommend Anime and Mangas in a Cold-Start Scenario






No releases published


No packages published