Recommendation system

SUBMITTED BY
Pratistha Singh
RECOMMENDATION SYSTEM

CONTENTS :-
 INTRODUCTION
 PURPOSE OF RECOMMENDATION SYSTEM
 TYPES OF RECOMMENDATION SYSTEM
 HOW TO IMPLEMENT RECOMMENDATION
SYSTEM
 EXAMPLES
 MAINTENANCE
 FUTURE SCOPE
 CONCLUSION

PURPOSE OF RECOMMENDATION SYSTEM
 Retrieval perspective
 Reduce search costs
 Provide "correct" proposals
 Users know in advance what they want
 Prediction perspective
 Predict to what degree users like an item
 Most popular evaluation scenario in research
 Interaction perspective
 Give users a "good feeling"
 Educate users about the product domain
 Convince/persuade users - explain
 Conversion perspective
 Commercial situations
 Increase "hit", "clickthrough", "lookers to
bookers" rates
 Optimize sales margins and profit
R.S.

TYPES OF RECOMMENDATION SYSTEM
 CONTENT – BASED
 COLLABORATIVE BASED
 HYBRID RECOMMENDATION
SYSTEM
R.S.

 CONTENT – BASED RECOMMENDATION SYSTEM
 Recommend items similar to those users preferred in the past
 User profiling is the key
 Items/content usually denoted by keywords
 Matching “user preferences” with “item characteristics” works
for textual information
 Vector Space Model widely used
R.S.

 COLLABORATIVE BASED RECOMMENDATION
SYSTEM
 Use other users recommendations (ratings) to judge item’s utility
 Key is to find users/user groups whose interests match with the
current user
 Vector Space model widely used (directions of vectors are user
specified ratings)
 More users, more ratings: better results
 Can account for items dissimilar to the ones seen in the past too
R.S.

 HYBRID RECOMMENDATION SYSTEM
R.S.

HOW TO IMPLEMENT RECOMMENDATION
SYSTEM

MAINTENANCE
 Costly
 Information becomes outdated
 Information quantity (large, disk space expansion)

FUTURE SCOPE
 Extract implicit negative
ratings through the analysis of
returned item.
 How to integrate community
with recommendations
 Recommender systems will be
used in the future to predict
demand for products, enabling
earlier communication back
the supply chain.

CONCLUSION
 Recommending and personalization are
important approaches to combating
information over-load.
 Machine Learning is an important part
of systems for these tasks.
 Collaborative filtering has problems.
 Content-based methods address these
problems (but have problems of their
own).
 Integrating both is best.

Recommendation system

More Related Content

What's hot (20)

Similar to Recommendation system (20)

Recently uploaded (20)

Recommendation system