In the paper Deep Neural Networks for YouTube Recommendations, it mentions that the “example age” feature helps recommending fresh contents in Section 3.3.

Many hours worth of videos are uploaded each second to YouTube. Recommending this recently uploaded (“fresh”) content is extremely important for YouTube as a product.


The distribution of video popularity is highly non-stationary but the multinomial distribution over the corpus produced by our recommender will reflect the average watch likelihood in the training window of several weeks.

To correct for this, we feed the age of the training example as a feature during training. At serving time, this feature is set to zero (or slightly negative) to reflect that the model is making predictions at the very end of the training window.

What is exactly is the "age"? Is it end_of_window - upload_time?
Why is it set to zero at serving time? How does it help with the freshness of the recommendation system?


You are correct. The figure from the paper shows how "days till upload" affect the results, so age is the days count.

enter image description here

Authors seem to say that it works because the model learns how does the age of the content affect popularity (e.g. older couple content will have more likes, because more people had chance to watch and rate it, or older "old", so not interesting anymore). At serving time they set it to zero, so the model threats everything as if it was a newly uploaded content, without the correction for age.

  • $\begingroup$ It is difficult for me to understand the reasoning "so the model threats everything as if it was a newly uploaded content". At serving time, why not just feeding the age of the video? Why does the model need to treat the videos as newly uploaded contents at serving time? $\endgroup$ – leon Feb 7 at 13:18
  • $\begingroup$ @leon you should refer to the paper for details. $\endgroup$ – Tim Feb 7 at 13:24

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.