omnisuggest: a ubiquitous cloud-based context-aware recommendation system for mobile social networks

22
OMNISUGGEST CONTEXT-AWARE RECOMMENDATION SYSTEM FOR MOBILE SOCIAL NETWORKS Submitted By: JOSHWA PHILIP B-TECH, CSE VJEC, KERALA 27-02-2015 1

Upload: joshwa-philip

Post on 20-Jul-2015

51 views

Category:

Technology


1 download

TRANSCRIPT

OMNISUGGEST

CONTEXT-AWARE RECOMMENDATION SYSTEM FOR

MOBILE SOCIAL NETWORKS

Submitted By:

JOSHWA PHILIP

B-TECH, CSE

VJEC, KERALA

27-02-2015 1

INTRODUCTION

• Retrieval to the filtering of relevant information.

• Relevant personalized information.

• Diverse and overloaded sources of information.

• Users’ historical and contextual data that best match the user’s preferences.

• E.g.: Facebook, Amazon, Foursquare

• Venue recommendation problem; solution using Collaborative Filtering(CF).

27-02-2015 2

PROBLEMS OF EXISTING SYSTEM

• Data sparseness

Sparse user-venue check-in matrix

• Cold start

The cold start problem in many existing Collaborative Filtering(CF)

recommendation systems

• Scalability

Reduced dataset size and recommendation quality.

27-02-2015 3

CONTRIBUTIONS IN OMNISUGGEST FRAMEWORK

• Recommendation framework combines social computing, and

recommendation modules, on cloud infrastructure

• Utilizing HITS (Hyperlink-Induced Topic Search) method

• Combination of Collaborative Filtering (CF) and group satisfaction

principle.

27-02-2015 4

WHAT IS RECOMMENDATION SYSTEM ?!

27-02-2015 5

• Information filtering system that seek to predict the 'rating' or

'preference' that user would give to an item.

EXAMPLE FOR OMNISUGGEST FRAMEWORK

27-02-2015 6

SYSTEM ARCHITECTURE

Major Components

User profiles.

User identification, venue name and identification, venue location (GPS

location, city, and country), time at which user performed check-in at a

venue.

Top-K Users and Venues.

Distribution and parallel execution of processing tasks on cloud

framework as each place has local sets of users and venues.

27-02-2015 7

Recommendation Module.

Ant colony algorithm and collaborative filtering (CF) is then

applied to generate an optimal solution in the form of venues that

best match an active user’s preference.

27-02-2015 8

Cloud Services Mapping

OmniSuggest framework follows a SaaS approach through a modular

service based architecture.

The SaaS forms the top layer of the cloud stack, offering real-time

personalized recommendations to a user or groups of users, while

abstracting underlying implementation details

Users access the service using thin clients, such as mobile devices, and

are typically unaware of the physical locale of the hosted service.

27-02-2015 9

27-02-2015 10

PROPOSED RECOMMENDATION FRAMEWORK

• An offline processing module

popularity ranking of users and venues

similarity graph creation among popular users

• An online recommendation module

• Group Recommendation

• Time Complexity

27-02-2015 11

AN OFFLINE PROCESSING MODULE

User Venue Popularity Ranking

Popularity ranking to users and venues for various category

hierarchies in a geographic location.

The HITS mechanism is utilized to perform the ranking for

producing a set of experienced users and popular venues.

27-02-2015 12

Hubs Similarity Graph Creation

This phase creates similarity graphs among experienced users (hubs)

under the various predefined categories.

The idea is to generate a network of like-minded people who share the

similar preferences for various venues they visit in a geographical

region.

27-02-2015 13

27-02-2015 14

Fig.4.

• (a) Hubs similarity graph retrieved from

database.

• (b) Connectivity of active user with hub

similarity graph.

ONLINE RECOMMENDATION FOR SINGLE USER

• Online recommendation framework that applies a variant of the Ant

colony approach

• On a graph of experienced users (hubs) to generate a set of the most

popular venues not previously visited by an active user.

• Most of the popular collaborative filtering techniques used.

27-02-2015 15

27-02-2015 16

GROUP RECOMMENDATION

• The existing systems focuses on recommending venues to individual

users based on personal preferences.

• Challenge: system must provide recommendations to a group of friends.

• produce recommendations that satisfy every group member, as an individual’s

context (e.g., speed, distance, and road traffic conditions) may vary with time.

27-02-2015 17

27-02-2015 18

GROUP RECOMMENDATION

• Initializations.

• Real-time processing for each member.

• Average, Least Misery, Most Pleasure, and Approval Voting

• Venue recommendation based on group satisfaction.

27-02-2015 19

CONCLUSION

• Cloud based solutions for the venue recommendation problem in social networks

for a single user and/or a group of friends.

• OmniSuggest framework considers:

a) the collective opinions of the experienced users

b) the effect of dynamic real-world physical factors(person’s distance from venues, speed,

weather conditions, and travel conditions)

• The scalability issues were addressed by proposing a cloud-based architecture that

allocated data and computational load on geographically distributed cloud nodes.

27-02-2015 20

BIBLIOGRAPHY

• http://ieeexplore.ieee.org/xpl/login.jsp?tp=&arnumber=6678350&url=http%3A%2F%

2Fieeexplore.ieee.org

• http://en.wikipedia.org/wiki/Cloud_storage

• http://en.wikipedia.org/wiki/Media_Cloud

27-02-2015 21

27-02-2015 22