3. INTRODUCTION
• The goal of RS is to produce a meaningful recommendation to a user for items or products on
the basis of their interesor preferences.
• Recommendation of books on Amazon, movies on Netflix or songs on Spotify is the real
world example of RS.
• Recommender systems were designed to fulfill the gap between the information and analysis
by filtering all the available information to provide the most usable information to the user.
• On the basis of user profile, RS has ability to predict whether a user prefers an item or not.
4. LITERATURE SURVEY
Title of Paper Conclusion
1)A Case-Based Recommendation
Approach for Market Basket Data.
After compared the performance of
developed RS
conclude that CBR is the good method in
case of
transactions.
2)RecommenderSystems: An overview
of different approaches to
recommend actions
The three approaches of recommendation
system
and their advantages and disadvantages.
3)Recommender Systems
Handbook.Springer.
In the unique approaches,hybrid robust
filtering
methods are better.
4)Towards privacy in a context-aware
social
network based recommendation
system
Focus on protecting data and request for
data, at
the point of data collection.
6. ARCHITECTURE
As per users feedback datastore in database.
▪ Then recommendation algorithm in which
data mining which describes thecollection of
analysis technicbuild recommendation model.
▪ Artificial intelligence predictthe users choice.
▪ Information retrieval addressthe information.
▪ Then recommendtion send to user.
7. ALGORITHM
• Recommendation algorithms can bedifferentiated
on the basis of knowledgesources that they
use.There are threetypes of knowledge source
which are usedfor classification recommenda-
tionalgorithms
• Social Knowledge: In this type,
socialrelationship of the user is used to
describethe recommendation algorithms like
tagsratings.
• Invidual Knowledge: Individual data ofthe user
is used to classify therecommendation algorithms
like behaviorand interest.
• Content Knowledge: Features of itemsare used
to differentiate therecommendation algorithms
8. RECOMMENDATION SYSTEM & ISSUES
1) Personalized Recommendation System:
Analyze the present & pas tbehaviour of user
send
recommendation.
▪ Content Based filtering
▪ Collaborative filtering
▪ Hybrid filtering
▪ Demographic filtering
2)Non-Personalized Recommendation System
In the non personalized RS,
items are recommended to
users on the basis of other users
review on the items
9. CONTENT BASED FILTERING
➢ Recommend similar to thoseuser has liked
in past.
➢ Based on content of items.
10. COLLABORATIVE FILTERING
➢ User with similar interest have common
preferences.
➢ Sufficiently large number of user
preferences are available.
11. USER BASED AND ITEM BASED
1. User Based
▪ In this same taste users are kept in the same group.
▪ In this approach recommendation are given to user based appraisal of
item by other users of thesame group
2.Item Based
▪ In the item based CF
algorithm, the similarities between different items are calculated in the dataset
▪ Once the most similar items are found, the recommendation is then computed by taking a
weighted average of the target user’s ratings on the similar items.
12. HYBRID RECOMMENDER SYSTEM
➢ Combine Collaborative filtering and
Content-based filtering.
➢Hybrid Recommender belongs to
application domain,the evolution process and
Proposed future research direction.
14. ISSUES
Cold Start Problem: The cold star problem arises when the new products are added in tot he catalogue or new user enters
into the system.
● Synonym: This problem arise when same item is represented with more different names or having similar meanings.
● Shilling Attacks: what happens if number of users and competitor giving the false ratings to some products either to
increase its popularity or to decrease into the RS
● Privacy: For the better recommendation to the user, RS need the personal information of the user but it may lead to
issues of user data privacy and security
● Scalability: The problem of scalability arises when the number of users of items increases tremendously. Computation
normally grows linearly with the number of usersand items.
● Grey Sheep Problem: Grey Sheep problem occur with people who do not completely agree or disagree with a group of
people
● Diversity: Primary factor on which most of the recommender system works upon is that similar user is likely to have
similar taste.
● Data Sparsity: This problem arises because the users generally rate only a limited number of products. There are
thousands of products available most of which are generally not used by the users and hence, they are not rated by the
users
15. APPLICATION AND FUTURE SCOPE
Application
▪ E-commerce site
▪ Content for website
▪ Entertainment
▪ Services
▪ YouTube
▪ Netflix
▪ Amazon
Future Scope
• Furthermore, this ppt also envisions the future of RS
which may open up new research directions in this
domain.
16. • ADVANTAGES
• Deliver Relevant Content
• Engage Shoppers
• Increase Number of Items per Order
• Reduce Workload and Overhead
• Offer Advice and Direction
• DISADVANTAGES
• Require large amount of data
• Changing Data
• Changing User Preferences
17. CONCLUSION REFERENCE
• The Internet is a basic source of
information nowadays where a large
amount of data is stored. RS helps the user
to find the require data from the web with
less effort, less spending time with more
accuracy.
• In this paper various recommendation
techniques and their related issues are
discussed. CB and CF are mostly used
techniques but they have some individual
problems like cold start problem synonyms
etc.
• https://en.m.wikipedia.org/wiki/Recommender_system
• https://www.researchgate.net/publication/338547981_
Recommendation System_Techniques_and_Issues