Personalized Difficulty Recommender for Foreign Language Texts

[mircealungu]

Zeeguu is a research project (read more about it) and a system that aims to investigate the impact of personalized reading for learners of a foreign language.

The system includes a difficulty estimator, but it is not personalized. That means, that the system has not achieved yet one of it's most important goals: of finding and recommending materials that are interesting and at the right difficulty level for learners.

The goal of this project is to devise a personalized difficulty estimator for zeeguu that estimates the difficulty of a text with respect to a given reader based on the following features:

• the words that the learner has translated in the past (meaning that they did not know them at that time)
• the words that they saw in vocabulary exercises (which are also part of the platform)
• user feedback on the difficulty of texts (too easy, ok, difficult)

Note that the difficulty estimation has to be integrated in the bigger picture of the content recommender that needs to find also materials that are interesting to the reader. And that the difficulty recommender does not have to be very precise, but just good enough because:

• if the text is a bit on the easy side no reader will complain
• if the text is a bit hard, the platform supports the learner with easy in-text translations so it's also fine

However, the recommender should definitely avoid giving too difficult texts to the learner because that would likely demoralize them!

[dawidwoz]

I believe this website has been closed, so the project rather does not make sense anymore.

[mircealungu]

Thanks @dawidwoz for spotting the problem! The website has simply moved to zeeguu.org. I've updated the url correspondingly!