Scribd
Upload
11 views20 pages

Recommender Systems

Recommender systems are designed to suggest items to users based on various data types, commonly used by platforms like Netflix and Amazon. Techniques include manual curation, recommending popular items, user-based collaborative filtering, and matrix factorization, each with its own advantages and limitations. User-based collaborative filtering identifies similar users to suggest items, while matrix factorization uncovers latent features from user-item interactions.

Uploaded by

abiahmoljoy
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
11 views20 pages

Recommender Systems

Recommender systems are designed to suggest items to users based on various data types, commonly used by platforms like Netflix and Amazon. Techniques include manual curation, recommending popular items, user-based collaborative filtering, and matrix factorization, each with its own advantages and limitations. User-based collaborative filtering identifies similar users to suggest items, while matrix factorization uncovers latent features from user-item interactions.

Uploaded by

abiahmoljoy
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 20

Recommender

Systems
Manual Curation, Recommending What’s Popular, User-Based
Collaborative Filtering, Item Based Collaborative Filtering, Matrix
Factorization.
• Recommender systems are tools designed to suggest items to users
based on various types of data.
• Commonly, they are used in platforms like Netflix, Amazon, and
Twitter to recommend movies, products, or users to follow,
respectively.
• Recommendations
Manual Curation
• Before the digital age, recommendations were often given by experts
such as librarians, who suggested books based on a person's interests
and previous reads.
• This personal touch ensured that recommendations were tailored and
relevant.
Recommending What’s Popular
• One straightforward approach to making recommendations is to
suggest popular items.
• This method is simple and can be effective, especially for new users
without much activity history.
Calculate Popular Interests:
• Use a Counter to count the occurrences of each interest across all
users.
Suggest Popular Interests:
• Create a function to suggest the most popular interests that a user is
not already interested in:
User-Based Collaborative Filtering
• User-based collaborative filtering makes recommendations by
identifying users who have similar interests and suggesting items
those similar users like.
1. Cosine Similarity Function:
2. Creating Interest Vectors:
• First, compile a list of unique interests:
3. User Interest Vector:
• Create a vector for each user where each element is 1 if the user has
that interest, otherwise 0:
4. User Interest Matrix:
• Generate the user interest matrix:
5. Compute User Similarities:
Calculate pairwise cosine similarities between users:
Most Similar Users:
• Function to find the most similar users:
7. Generating Recommendations:
• Suggest new interests based on similar users’ interests:
• Curse of Dimensionality: In high-dimensional spaces, most vectors
are far apart, which can reduce the effectiveness of similarity
measures.
• Scalability: User-based collaborative filtering can struggle with large
datasets.
• Advantages: Simple to implement, effective for small datasets.
• Limitations: Less effective for large datasets, susceptible to sparse
data problems.
Matrix Factorization
• Matrix factorization is a technique used to uncover latent features
from observed data.
• In the context of recommender systems, it is often used to
decompose a user-item interaction matrix into two lower-dimensional
matrices: one representing users and the other representing items.
• GO THROUGH THE TEXT BOOK ALSO

You might also like