This document discusses how to build a personalized news recommendation platform. It explains that recommendation systems are needed to retain users, increase traffic, and improve the content experience. It describes popular techniques like collaborative filtering, content-based filtering, and hybrid systems. Specifically, it outlines a case study using a USPA framework with real social news data. Key factors for a news recommendation system are discussed like novelty, user history, and location. The document also provides a simple example of building a recommendation engine with Apache Spark.
How to build a Personalized News Recommendation Platform
1. How to build a Personalized News
Recommendation Platform
By Nguyễn Tấn Triều (Thomas)
Contact Email: tantrieuf31@gmail.com
BigDataVietnam.org
2. Agenda
1. Why do we need recommendation systems ?
2. How can we design recommendation systems ?
3. Case study: News Recommendation with USPA framework
11. Why Should We Use Recommendation Engines?
38% of click-through rates on Google News are
recommended links
12. The value of recommendation system
1. Retain user loyalty
2. Builds the volume of user traffic
3. Delivers best content experience to reader
4. Give your business a wider marketplace
13. So what is recommendation engine ?
In technical terms, a recommendation engine problem is to develop a
mathematical model or objective function which can predict how
much a user will like an item.
If U = {users}, I = {items} then F = Objective function and measures the
usefulness of item I to user U, given by: F: U x I → R
Where R = {recommended items}.
For each user u, we want to choose the item i that maximizes the
objective function:
15. Popular techniques to build recommendation systems
1. Collaborative Filtering
2. Content-Based Filtering
3. Hybrid Recommendation Systems with USPA framework
16.
17. User-based Collaborative Filtering
● Based on a large amount of information on users’ behaviors, activities or
preferences
● Predicting what users will like based on their similarity to other users.
● A key advantage:
○ accurately recommending complex items such as news without
requiring an “understanding” of the item itself
19. Content Based Filtering
( Item-based collaborative filtering)
Key ideas:
● Based on a description of the item and a profile of the user’s preference.
● Keywords are used to describe the items
● User profile is built to indicate the type of item this user likes.
21. Hybrid Recommendation with USPA framework
These methods can be used to overcome some
of the common problems:
1. cold start (no information on users’ behaviors)
2. sparsity problem (dense matrix)
27. Notes for designing news recommendation system
1. Novelty – In general, users tend to be more interested in the latest news
rather than in something that happened a long time ago.
2. User history – The latest news a user has read are very important to produce
recommendations, because the user is intentionally showing interest on a topic
or a set of topics.
3. Location – Users are more interested in news related to nearby events: the
closer a user is to the place of the happening, more probably this can affect
him. A news recommender system should then take into account the location
where the action described in the piece of news took place. In a mobile
environment scenario, user location has to be frequently updated and
considered.