Data Science Stack Exchange
Data Science Stack Exchange

Questions tagged [recommender-system]

Everything related to recommender systems

Everything related to recommender systems

Types of Recommender Systems Some types of recommender systems:

  1. Collaborative Filtering

    • Based on user-interaction data such as "user A watches movie Z"
    • Common methods: Matrix Factorization, Neighborhood Models

  2. Content Based Recommender Systems

    • Based on user and item content such as user demographics, movie genres, actors in the movie, etc
    • Common methods: Classification, Regression
483 questions
10
votes
1 answer

How should one deal with implicit data in recommendation

A recommendation system keeps a log of what recommendations have been made to a particular user and whether that user accepts the recommendation. It's like user_id item_id result 1 4 1 1 7 -1 5 19 1 5 80 …
wdg
  • 203
  • 1
  • 6
4
votes
1 answer

Why softmax in YouTube’s DNN recommender

I am confused about the softmax layer of YouTube’s DNN candidate generation. A user may interact with many videos. Softmax is assuming classes are exclusive. For example, logits = [[4.0, 4.0, 1.0]], labels = [[1.0, 1.0, 0.0]], the sigmoid cross…
lun yu
  • 43
  • 3
4
votes
2 answers

User-to-item and item-to-user recommendations

I'm currently creating a recommender system and there are different types of the systems. Does anyone know something about the user-to-item and item-to-user recommendations? I have been searching for days, but I couldn't find anything about them. I…
ayiram
  • 149
  • 5
4
votes
2 answers

Price optimization for tiered and seasonal products

Assuming I can collect the demand of the purchase of a certain product that are of different market tiers. Example: Product A is low end goods. Product B is another low end goods. Product C and D are middle-tier goods and product E and F are…
Guo Hong Lim
  • 143
  • 3
4
votes
2 answers

Evaluating recommendations quality and accuracy

I'm developing a recommendation system, that should provide my clients what actions they should take in order to hit certain targets. The underlying mechanics of the process is physical - where both actions and outcomes can be measured physically.…
yoav_aaa
  • 993
  • 4
  • 13
3
votes
1 answer

What is the "matrix trick" in recommendation systems?

I just found slides from Matt Gormley (CMU) about recommendation systems. Under the heading "Unconstrained Matrix Factorization" he mentions: Optimization problem SGD SGD with Regularization Alternating Least Squares User/item bias terms (matrix…
Martin Thoma
  • 18,880
  • 35
  • 95
  • 169
3
votes
2 answers

What kind of data is not appropriate using CF to do recommendation?

I am currently working on a recommendation system for daily news. At first, I evaluated all the recommender algorithms and their corresponding settings (e.g., similarities, factorizers, ...etc) implemented in Mahout. Since we want to recommend daily…
shihpeng
  • 553
  • 1
  • 4
  • 8
3
votes
4 answers

Can a recommendation system be built without any user ratings?

I was planning to make an artwork recommendation system as a project by using the WikiArt open source dataset available on kaggle, I'm still looking for datasets which might already have user ratings but is it possible to make a recommender without…
Kartikay
  • 41
  • 1
  • 3
3
votes
2 answers

recommender systems : how to deal with items that change over time?

Let's say I am building a recommender system where items change through time. We suppose that each transaction is composed of : an item $i$ in list of items $(i_1, i_2, i_3, .., i_m)$. a user $u$ in list of users $(u_1, u_2, u_3, ..., u_n)$. a…
Theudbald
  • 1,068
  • 8
  • 16
3
votes
1 answer

Use of negative correlation coefficient in pearson correlation algorithm for recommender systems

I am new to recommender systems and am trying to find similar users of base users for user-based collaborative filtering. When I calculated the similarity score between two users (based on their ratings with Pearson algorithm [or Resnick's weighted…
5im
  • 131
  • 2
2
votes
2 answers

In a SVD with user/video bias, why is the UV contribute so small?

I'm testing a SVD-based collaborative filter on my data set, in which the label, $r_{ij}$, is a real value from 0 to 1. Like the many papers suggested, to have a better performance, instead of using $ \hat{R} = U \cdot V^T $ directly, I use $\hat{R}…
ice_lin
  • 147
  • 6
2
votes
1 answer

How to use testing data set to measure recommender system algorithm

I am new to recommender systems and am trying to build one using item-to-time CF. Currently, I am trying to evaluate/measure results using MAE. I have one step which is unclear (after I managed to split training dataset and testing data set and…
Zhenkai
  • 121
  • 3
2
votes
1 answer

Advantages of Binary Rating System for Collaborative Filtering Recommender Systems

I notice that Netflix, which I think used to use a five-star scale for rating content and give predicted ratings for unrated content on the same scale, now just has basic like/dislike buttons. Music streaming services like Pandora and Spotify seem…
AffableAmbler
  • 363
  • 1
  • 2
  • 11
2
votes
2 answers

Which recommender system approach allows for inclusion of user profile?

I wanted to enhance a recommendation engine with information relying not only on past purchases or ratings but also on behavioral and demographical variables like sex, age, location, service usage frequency or hours. This information may be sparse…
user2530062
  • 297
  • 2
  • 8
2
votes
0 answers

Taxonomy of recommender system methodologies

There's tons of material online but yet I can't reconcile the different definitions for recommender system methodologies / strategies. I think we can identify several axes: memory vs model based; Model based models do not need to store historical…
pincopallino
  • 73
  • 5
1
2 3 4