New answers tagged

1

Since you have access to user level features, try a k nearest neighbors recommender algo. When a user signs into your site, find the k most similar looking users (based on a similarity metric such as cosine similarity), and recommend products that those users have purchased.


1

I now have a working network. It turned out that the gradients were all zeros after only about 3000 update step. I tried two approaches to fix this - using Batch Normalization after each activation function in the feed-forward net and changing the activation function from ReLU to Leaky ReLU. Both worked, and I ended up using the Leaky ReLU without ...


1

When you have some historical data on good/bad matches (and "okay" features to describe these matches), you can try a Siamese Neural Network. This type of model is a "few shot" model, meaning that it is designed to work with a relatively small ammount of (training) data and potentially noisy features. Essentially you fit a model to pairs ...


1

Create a histogram of the watch_time_ms. If you are lucky - you may see a bi-modal distribution (i.e two peaks). The higher/lower peak could be interpreted as interested / uninterested behavior respectively. Then your threshold could lie somewhere in the valley between your two peaks in the histogram. If your videos are variable length - you may also want to ...


0

This is where your domain expertise comes into play. Some of the important considerations, e.g. location-based user behaviors and location-based offering of products. For a typical recommendation problem, all you need is the product ratings from the users. The product characteristics and users' behavior towards them are inferred. What you additionally have ...


0

You should consider it: Using a collaborative filtering model will implicitly learn the users similarity. If you extract the lower dimension representation of users this will give you what can be considered as user embeddings. If you perform clustering on this you can get user clusters. The issue is that unless you perform collaborative filtering with side ...


1

This is a kind of confirmation bias, it's often found in recommender systems and it's hard to beat: since the system is designed to find similar books, it's normal that it returns the most similar books but it can be disappointing for the user. This bias is especially strong if there are few past books to refer to, i.e. the reference sample used by the model ...


1

That's a good question. This task would be referred to as multi-label encoding. Bascially if a movie belongs to several genres, you one hot-encode each genre and add the vectors. If there were only 6 genres (horror, romance, action, adventure, comedy, fantasy) For instance a movie that is horror, action and comedy (The Dead don't Die?): horror = [1, 0, 0, 0,...


0

A quick answer. Note: Checking the references I could access fully, there are no discrepancies between the definitions as long as the terms are translated correctly. Some definitions: Relevant items: Are items the user(s) have themselves established as relevant with their actions. Recommended items: Are items which the recomendation system predicts will be ...


0

Making predictions for new users is often called the cold start problem. This problem is difficult to over come in a purely item-to-item recommendation system. The most common solution is to include more than just item interaction information. Examples include user profile information and item content information.


Top 50 recent answers are included