Loading presentation...

Present Remotely

Send the link below via email or IM

Copy

Present to your audience

Start remote presentation

  • Invited audience members will follow you as you navigate and present
  • People invited to a presentation do not need a Prezi account
  • This link expires 10 minutes after you close the presentation
  • A maximum of 30 users can follow your presentation
  • Learn more about this feature in our knowledge base article

Do you really want to delete this prezi?

Neither you, nor the coeditors you shared it with will be able to recover it again.

DeleteCancel

Make your likes visible on Facebook?

Connect your Facebook account to Prezi and let your likes appear on your timeline.
You can change this under Settings & Account at any time.

No, thanks

CONTENT-BASED FILTERING FOR MUSIC RECOMMENDATION

No description
by

Svetomir Kurtev

on 25 June 2015

Comments (0)

Please log in to add your comment.

Report abuse

Transcript of CONTENT-BASED FILTERING FOR MUSIC RECOMMENDATION

CONTENT-BASED FILTERING FOR MUSIC RECOMMENDATION
Implementation
Ivan Kodzhastoyanov
Georgi Vasov
Svetomir Kurtev

Agenda:

Problem
Existing solutions
Motivation
Technology
Approaches
Plan
Implementation
Memory-based approach
Model-based approach
Results
Conclusion
Problem
Plan
What is the problem ?
The current state of the music industry
Our motivation
Existing solutions
System Architecture
Memory-based Approach
Project restrictions
Too many choices to choose from
Selecting the right type of music
Information overload

Limited time
Insufficient competence & skills
GrooveShark ~ 25 million songs

Google Play Music ~ 20 million songs

Spotify ~ 20 million songs
Explosive growth of digital music libraries
and music streaming services
Last.fm
Pandora Internet Radio
General pattern of the solutions
Music library

User profile

An agent/subsystem for analysing and selecting music items
What do they have in common?
Conclusion
Thank you for your attention!
Approaching the problem as:


Music listeners (consumers)

Developers
Project Plan
Iteration 1

Focus on the implementation of the memory-based approach



Iteration 2

Focus on the implementation of the model-based approach
Comparison between the two approaches
Focusing on the creation of a recommendation system for music items

The system should work in a content-based manner

Collaborative Filtering would not be considered for the scope of the project

External sources would be used for analyzing the dataset and extracting the acoustic attributes of the songs
Technology
Content-based recommendation systems
Collaborative Filtering systems
Differences between the two approaches
Each content-based system contains the following core components:

Content Analyzer

Profile Learner

Filtering Component


A content based recommendation system can be build by using either a memory-based and/or model-based approach.
"
Collaborative Filtering is the process of filtering or evaluating items using the opinions of other people
"
Schafer & Frankowski
Basic premises for Collaborative Filtering:

Items receive ratings

If users had similar taste in the past, they will have similar taste in the future as well


Collaborative Filtering systems can also be build using memory- and/or model-based approaches.
User Interface (UI)
Filtering Component
Profile Learner
Recommender System (RS)
Content Analyzer
Acoustic attributes
High relevance
Low relevance
Energy
Danceability
Tempo
Valence
Liveliness
Speechiness
Loudness
Similarity measures

Euclidean distance
K-Nearest Neighbours Algorithm
Memory-based Approach
Results
The system makes better recommendations when using higher k-NN values.
Model-based Approach
Bayesian networks
Naïve Bayesian Classifiers
"Preference Model Algorithm"
* This approach makes use of the acoustic attributes as well
Model-based Approach
Results
The average results peak at 85% for α=0.15

None of the average values falls under 80%.
Comparison of approaches
Quality of recommendations
Time complexity
Space efficiency
Difficulties
Project writing methodology

Unfamiliar field

Pacing issues

Future work
Conclusion
Did we manage to accomplish the goals set in the problem statement ?
What can be added or improved further?
Addition of a music player

Addition of a User Interface (UI)

Conducting additional tests with different setups

Addition of more features such environment,mood etc.

How successful is the implementation ?
Are we satisfied with the results ?
Full transcript