Sie sind auf Seite 1von 25

STES’s

Smt. Kashibai Navale College of Engineering


Department of Computer Engineering
Year 2017-18
SEM II

PARS* : Preference Aware


Recommender System
Group Member :
 Group Id : 20 Vijay Galande
Amol Kedar
 Guide name:M.R.PATIL Kartik Fating
Krushna Kotgire
OUTLINE
1. MOTIVATION
2. INTRODUCTION
3. PROBLEM STATEMENT
4. ABSTRACT
5. SCOPE
6. LITERATURE SURVEY
7. GAP ANALYSIS
8. PROPOSED ARCHITECTURE
9. UML DIAGRAMS
10. ALGORITHM USED
11. SOFTWARE-HARDWARE REQUIREMENTS
12. PLAN OF PROJECT EXECUTION
13. ADVANTAGES
14. CONCLUSION
15. FUTURE SCOPE
16. REFERENCES
 MOTIVATION
Recommendation systems help users find and select items (food, restaurants) from the
huge number available on the web or in other electronic information sources . Given a
large set of items and a description of the user's needs, they present to the user a small
set of the items that are well suited to the description. Recent work in recommendation
systems includes intelligent aides for filtering and choosing web sites, news stories , TV
listings, and other information.
As a result, PARSE it seems desirable to have personalized intelligent systems that process,
filter, and display available information in a manner that suits each individual using them.
The need for personalization has led to the development of systems that adapt
themselves by changing their behavior based on the inferred characteristics of the user.
 INTRODUCTION
Recomender systems make use of community opinions to help users identify useful items from
a considerably large search space (e.g., Amazon inventory, Netflix movies). The technique
used by many of these systems is collaborative filtering (CF), which analyzes past community
opinions to find correlations of similar users and items to suggest k personalized items (e.g.,
movies, restaurants) to a querying user
Our system is a preference aware recommender system built specifically to produce high-
quality recommendations in an efficient manner. It produces recommendations using a
taxonomy of three types of ratings :(1) Spatial ratings for non-spatial items, represented as a
five-tuple (user, ulocation, rating, item, ureviews), where ulocation represents a user location,
for example, a user located at home rating a book; (2) non-spatial ratings for spatial items,
represented as a four-tuple (user, rating, item, ilocation) , where ilocation represents an item
location , for example ,a user with unknown location rating a restaurant ;(3) spatial ratings for
spatial items , represented as a five-tuple (user, ulocation , rating , item , ilocation) , for
example, a user at his/her office rating a restaurant visited for dinner.
 PROBLEM DEFINITION

Preference locality suggests users from a spatial region (e.g., neighborhood) prefer items
(e.g., hotel, area) that are manifestly different than items preferred by users from other, even
adjacent, regions. Preference locality suggests that recommendations should be influenced by
review based spatially close to the user. The intuition is that localization influences
recommendation using the unique preferences found within the spatial region containing the
user.
when recommended items are spatial, users tend to travel a limited distance when visiting
venues. We refer to this property as “travel locality.” In analysis of foursquare data, it was
observed that 45% of users travel 10 miles or less, while 75% travel 50 miles or less.
 ABSTRACT
Preference aware recommender system that uses location-based ratings to produce
recommendations. Traditional recommender systems do not consider spatial properties of
users nor items, PARS, on the other hand, supports a taxonomy of three novel classes of
location-based ratings, namely, spatial ratings for non-spatial items, non-spatial ratings for
spatial items, and spatial ratings for spatial items.

PARS exploits user rating locations through user partitioning, a technique that influences
recommendations with ratings spatially close to querying users in a manner that maximizes
system scalability while not sacrificing recommendation quality. PARS exploits item
locations using travel penalty, a technique that favors recommendation candidates closer in
travel distance to querying users in a way that avoids exhaustive access to all spatial items.
PARS is efficient, scalable, and capable of producing recommendations twice as accurate
compared to existing recommendation approaches.
 SCOPE

Online “Preference Aware Recommender System” will be a mobile and website


based application which will be based on data mining, text mining and
recommendation system. Its main aim is to provide accurate recommendations
to users based on spatial and non- spatial properties of the user and item
requested. Item maybe a certain food item or a certain restaurant providing a
type of item.

Users may be able to review both restaurants and items. The users reviews would
be examined to create a profile of the user. The product will consider the users
history (from the previous likings and reviews) and/or location before providing a
recommendation. The application will take the users search request and provide
the relevant recommendation
 LITERATURE REVIEW
Number Publication Technique Used Scope For Improvement Accuracy( Scalability
%)
R-1 IEEE, 2016 Location, Time, and Traditional recommender 72 high
Preference used. systems do not consider
spatial properties of users
nor items
R-2 IEEE, 2015 Location is used in this It has less accuracy. 62 low
system.

R-3 IEEE, 2016 A Multi Criteria Review is Technique that influences 74 high
used. recommendations not
sacrificing
recommendation quality.

R-4 IEEE, 2014 Item-to-Item Collaborative It has less accuracy. NA high


Filtering is used.

R-5 IEEE, 2015 Personalized Location is It has more accuracy; but 69 high
used. less recommendation rate.
8
 GAP ANALYSIS
Sr. Title Publication Gap Analysis
No.
1. Location, Time, and Preference Aware IEEE, 2016 By using Mopnar,Porter stemming
Restaurant Recommendation Method. algorithm error rates will be
reduced due to inherent nature of
the algorithm.
2. LARS*: An Efficient and IEEE, 2015 Porter stemming algorithm is
Scalable Location-Aware Recommender advance version of
System recommendation algorithms.
3. A Multi Criteria Review-Based Hotel IEEE, 2016 Mopnar,Porter stemming
Recommendation system. algorithm has more accuracy and
recommendation rate.

4. Amazon.com Recommendations Item-to- IEEE, 2014 By using Porter stemming


Item Collaborative filtering. algorithm it increases accuracy
and recommendation rate.

5. City Recommendation Based on User IEEE, 2015 Porter stemming algorithm has
Communities, ”Web Congress more accuracy and
recommendation rate.
 ARCHITECTURE
1. Users:
The same application may be used by several users. In addition, users that
interact with the host application may be either registered or anonymous
and may interact frequently or occasionally. This impacts the architecture
of the recommender (requires different identification means, e.g.
comments, preference).

2. Application infrastructure:
The infrastructure the host application runs on puts strong constraints on the
types of recommendation algorithms that can be used and on their
specific implementation. In particular, the scalability of the solution has to
be carefully studied. Two main cases can be identified, whether the
application is accessed through a browser or if an application runs locally
on the user device.

System: Recommendation percentage is


calculated based on users comments by Collaborative filtering
algorithm. It could be on the basis of service, cost, test or
ambiance.
 UML
DIAGRAMS

Use Case Diagram


Activity Diagram
 SEQUENCE DIAGRAM

Department of Computer Engg. SKNCOE 7/29/2019


 PORTER STEMMER ALGORITHM
The porter stemming algorithm is a process for removing suffixes from words in
English . Removing suffixes automatically is an operation which is especially useful
in the field of information retrieval. In a typical information retrieval environment a
document is represented by a vector of words .

Usually is desired that only inflectional morphemes are removed (those


corresponding to declination , conjugations , etc), not also derivational
morphemes ( which correspond to different parts of speech).
Input : A Word = Stem + Affix(es).
Output : stem .

 Steps:
 1) The words with common meaning are selected For eg : Connected ,
Connection.
 2) The performance of IR system will be improved if term groups are
converted into single term eg: Connect.
 3)this is done by removal of various suffixes –ion, -ed, -Ing .
 4)in addition suffix stripping process will reduce total number of terms in IR
system and hence reduce the size and complexity of the data.
 MOPNAR ALGORITHM

MOPNAR is an extension of multi-objective evolutionary algorithm (MOEA). It


helps in mining with a low computational cost a reduced set of positive and
negative words that are easy to understand and have good tradeoff
between the number of rules, support, and coverage of the dataset.
The main focus of the algorithm is to obtain a reduced set of words which are
having good tradeoff considering three objectives which are
comprehensibility, interestigness and performance. In order to perform a
learning of rules it extends the traditional MOEA model.
.
Input : N: number of sentences in the form of paragraph; x={x1,x2,x3…xN}.

Output: (1) Positive and Negative sentences are separated.


(2) Positive and Negative sentences count.

Steps:
1) Upload input data file containing ‘n’ number of paragraphs.
2)Data will be validated using sentiment sentences validation.
3)Sentences will be segmented.
4)Tokens will be separated from all sentences.
5)Process of segregation is performed for finding the positive, negative words
by matching the words with POS tagging words.
6)Positive and Negative sentences and their counts are stored.
 SOFTWARE-HARDWARE REQUIRTMENTS

Technology- JAVA

Database- SQL / Mongodb/(Casendra)

UI- HTML
CSS
JQuery
Bootstrap
Ajax

Algorithm- Mopnar algorithm


Porter stemming algorithm
Selection algorithm

Tools- Eclipse
SQL
PLAN OF PROJECT EXECUTION
 ADVANTAGES
 This system allows to search for a specific food item directly.
 Allows users to write reviews about that place and food.
 Each and every review is processed for both user and restaurants.
 Ratings depend on the user/hotel review count which is distributed amongst five
parameters.
 Along with user details it has an extra-ordinary feature of graphically representing
the users liking based on the reviews written by the user.
 Graphical bar chart display representation of restaurant parameters of Taste,
Service, Ambience, Cost which is calculated on reviews.
 This recommender system is recommend nearby restaurant.
 This was helpful for the user who is new in that city like tourist.
 It allows users to add a food item which they liked at restaurant while writing their
review.
 CONCLUSION

To achieve our goal, we have divided the calculation process to calculate the
recommendation process for each user. Thus producing a more personalized
search result for a user. These recommendations are generated through the
personal preferences of user.
The user also gets recommendations for a specific item with the help of preference
or review based collaborative filtering. This help user to find places which are more
suitable to their preferences and comfort. This enables a person to have a more
comfortable and suitable dining experience and also sharing their reviews to help
others in the same manner.
 FUTUTRE SCOPE
Letting users help other users with their reviews and liking for food, will automatically
increase the web to help in finding new and interesting food items. The ability to
find specific food items will certainly help users to find a place to eat, where it is
best served and their craving would be quenched. Alongside as the network of
new restaurants.
Considering the technical aspects, as the data will increase and the scalability of
the system would be under question, it will be very affordable and time efficient, as
our system has the capability to have shards (nodes).
Therefore all of the data will be distributed amongst these nodes, and there will be
no need to transfer data from the existing machine to a higher powered machine.
And a persons craving for a specific food item in their vicinity would be satisfied in
a very true sense. This will help in increasing business, as the system would be able
to provide the user with exactly what the user wants in majority of the cases, and
this will help in boosting business.
 REFERENCES

1. Md. Ahsan Habib, Md. AbdurRakib, and Muhammad AbulHasan. “ Location, Time, and Preference Aware
Restaurant Recommendation Method”19th International Conference on Computer and Information
Technology, December 18-20, 2016, North South University, Dhaka, Bangladesh.
2. Mohamed Sarwat , Justin J. Levandoski, Ahmed Eldawy and Mohamed F. Mokbel “ LARS*: An Efficient and
Scalable Location-Aware Recommender System”TRANSACTIONS ON KNOWLEDGE AND DATA
ENGINEERING, VOL. 6, NO. 1, NOVEMBER 2015.
3. Yashvardhan Sharma, Jigar Bhatt, RachitMagon “ A Multi Criteria Review-Based Hotel Recommendation
System”2016 IEEE International Conference on Computer and Information Technology.
4. Greg Linden, Brent Smith, and Jeremy York “ Amazon.com Recommendations Item-to-Item Collaborative
filtering”2014 IEEE International Conference on Computer and Information Technology.
5. Bidart R, Periera A. C.M., Almeida J. M, Lacerda “ City Recommendation Based on User Communities,
”Web Congress (LA-WEB), 2015 9th Latin American, pp. 50 58.
THANK YOU

Das könnte Ihnen auch gefallen