Join for free and connect with our local tech scene

Stay on top of the latest companies and upcoming events with our weekly newsletter, and be counted among the people building the future of your local tech community.

Buffalo, NY /

Content based and Collaborative Filtering Recommendation Systems

Delaware North HQ 250 Delaware Ave , Buffalo, NY (map)

This will be an introductory talk on recommendation systems (RecSys)


The amount of information available to us in the information age is without a doubt a blessing, but the vast amount of choices made available quickly results in information overload. Without a way of ordering relevant information and filtering out the rest, decision making can become problematic, time consuming, and generally unpleasant. This is the problem of recommendation systems: how to prioritize relevant information to enable better decision making?

I will cover the two most commonly found systems: the Content Based and the Collaborative Filtering implementations. While aiming to solve the same problem, namely providing useful information to the user, each of these systems occupy a slightly overlapping solution space. For each kind of RecSys, I will cover how it solves the recommendation problem and where it is used. I will then cover the major functional components of the system, with example code of the major functional components. I will give more attention to Collaborative filtering as it is the more impactful of the two systems. I will go over measures of success and a side by side comparison of the two systems. Finally, I will introduce the concept of a hybrid RecSys which minimizes the weaknesses of each system.


This is a rough outline of what I plan to cover.

• What is a recommender system (RS)?

-- a.k.a RecSys, Recommendation Engine, (RE)

-- Where we find RS

• Motivating factors behind RS development

• The most commonly found RS

- Content Based (CB) Recommenders

-- Philosophy & Design Approach

-- Building a CB RS (code & math)

-- Example

- Collaborative Filtering (CF) Recommenders

-- Philosophy & Design Approach

--- Implicit vs Explicit Rating

-- Building a CF RS (code & math)

-- Methods of CF: Memory Based, Model Based

-- Example

- When is a RS considered successful?

-- Accuracy isn't everything

-- Target reveals pregnancy

- CB RS and CF RS: side by side comparisson

- Hybrid RS

• What to bring

Nothing planned at this time.

• Important to know

I plan to make this presentation accessible to as wide of an audience as possible. Those with little to no background in mathematics or machine learning should be able to follow along quite easily.

Submitted by

Eightbit-59902de7-98ba-42d1-a922-49a369f68e3b Brett Langdon


Sign in to comment.