Recommender systems have become an important part of various applications in e-commerce, supporting both customers and providers in their decision-making processes. However, these systems still must overcome limitations that reduce their performance, like recommendations overspecialization, less popular item providing, and difficulties when items with unequal probability distribution appear or recommendations for sets of items are asked. A novel approach, addressing the above issues through a case-based recommendation methodology is presented here. The scope of the presented approach is to generate meaningful recommendations based on items' co-occurring patterns and to provide more insight into customers' buying habits. In contrast to current recommendation techniques that recommend items based on users' ratings or history, and to most case-based item recommenders that evaluate items' similarities, the implemented recommender uses a hierarchical model for the items and searches for similar sets of items, in order to recommend those that are most likely to satisfy a user.
3. INTRUDUCTION
Recommendation System is a software tool
Provide meaningful and effective item recommendation to
the active user.
It support both User and providers
User- handling information overload and finding
adequate items to cover their needs
Provider-To identifying user needs and increasing the
amount diversity of the items they sell.
5. Literature Survey
1. Collaborative Filtering Recommender Systems
Items among those liked by similar users (“neighbors”) are
recommended to the active user.
A user profile is built of the items that the user has rated highly,
thus similarities in user tastes are deduced from previous ratings.
2. Content-based Recommender Systems: State of the Art
and Trends
• Recommend items similar to those a given user has liked in past.
• Matching up the attributes of a user profile in which preferences
and interests are stored, with the attributes of a content object
(item), in order to recommend to the user new interesting items.
6. 3. Mining Association Rules between Sets of Items in
Large Databases
Discover interesting hidden patterns and frequent
associations among existing items.
Applying new approach over voluminous sets of rules
in the post processing step, to reduce the number of
rules to several dozens or less.
4. Exploring Customer Preferences with Probabilistic
Topics Models
Capability to discover latent baskets and latent users
lays the foundation for understanding user tastes and
their relations to items
latent topic models can be effective in recommending
items to users even when applied to large data sets
and large item sets.
7. Methodology
Related Works
Case-based reasoning (CBR):
Problem solving System.
Uses old experience to solve new
problem.
“Similar problem have similar solution”.
Learn past/previous cases and give
solution to the new problem.
9. Presented work
Recommend the similar items of already
user has been selected some items.
Analyzing previous transaction items and
recommend similar item into the user.
Using case based recommendation reason
to identify and recommend the item.
• Every Transaction as a case and its items as
attributes.
11. Advantages
• Recommend the popular items to the user to
compare with the previous transaction of all
users.
• Recommend the meaningful and effective
item to the known user as well as new user.
• Saves the time of the user to search the
similar items.
12. Limitations
Not able to recommend the new items
because it recommend the item only on the
basis of previous transaction.
Need a lot of data to effectively make
recommendations.
13. Conclusions
Its ability to recommend complementary items to the
already selected ones.
Additionally, it can recommend popular items and
generate recommendations to new users.
Reducing the cold start and the overspecialization
problems of CF and CB techniques.
14. Future Work
Quality characteristics of the items and constraints related
to them could be incorporated (for example, to
recommend only new items).
It could be applied to other domains where the outcome
of a user’s experience depends on the total set of items
used together.
Finally we plan to examine the temporal characteristics of
the selected items.