A Collaborative Situation-Aware Scheme for Mobile Service Recommendation

Post on 30-Nov-2014

232 views 0 download

Tags:

description

Nowadays mobile devices, such as smartphones, are becoming increasingly popular and they offer a wide range of mobile applications, also called mobile apps, suitable for different situations. This abundance of applications makes the research over them difficult and time consuming. Context-aware resource recommendation for ubiquitous devices is aimed at proactively pushing personalized suggestions to mobile users, presenting them unseen mobile apps. Typically, the recommendation is based on recognizing the current situation of the user and suggesting them the appropriate resources for those situations. We believe that recommendation schemes can emerge from users’ collective behavior. An emergent behavior or emergent property can appear when a number of simple entities (agents) operate in an environment, forming more complex behaviors as a collective. In our case the entities are represented by mobile users who provide positional data through Global Positioning System (GPS) provided by almost all modern smartphones. The recognition task is performed by exploiting contextual information and preferably without using any explicit input from the user. Thus, in this thesis we present a collaborative multi-agent scheme for social events detection in which a stigmergic paradigm and fuzzy representations are employed to cope with the approximation typical of implicit and aggregated information. The multi-agent scheme is structured into three levels of information processing. The first level is based on a stigmergic paradigm, in which marking agents, following the mobile user, leave marks in the environment. The accumulation of such marks enables the second level, a fuzzy information granulation process, managed by event agents, in which relevant events can emerge. Finally, the third level, a fuzzy inference process, managed by situation agents deduces the user situation from the underlying events.

transcript

A Collaborative Situation-Aware Scheme for

Mobile Service Recommendation

CandidatoLuigi Massa Gallerano

RelatoriFrancesco MarcelloniBeatrice LazzeriniMario Giovanni C.A. Cimino

Smartphone Market

SmartphoneMarketStrong

Rise

A new report (Juniper Research) forecasts that the number of global smartphone shipments will reach one billion per annum in 2016

Smartphone Apps

Number ofSmartphone

Apps

Official Google Blog: “10 billion apps downloaded and installed as of December 2011”

Apps for each situation

Mobile Recommendation

App Recommendation

Context

Situation-awareness

This Autonomous Perception implies:

ReasoningDecision AdaptationCognitive system Intrinsic uncertainty in data

No Explicit InputFuzzy Logic

Emergent Paradigm

Stigmergy

Multi-Agent Scheme

UserSituation

Situation-awareness

Control to achieve results is distributed over all entities

The collaborative Multi-agent scheme

Marking Level

Fuzzy GranulationLevel

Fuzzy Inference

Level

Marking Agents

Event Agents

Situation Agent

Marks

Event Certainty Degree

Fuzzy Rules

EMERGES

UserSituation

Marking Level

Released every Tm seconds Intensity Spatial decreases (percentage δ per cell) Temporal decay (every Td sec of a percentage α) Superimpose

Movement Grid

Max Intensity User Still

Fuzzy Granulation Level

Grouping Agent

Disjoining Agent

Observes a Neighboring Area Calculates Intensity associated with the area

Computes Membership Function

Marking AgentMarking Agent

Marking Agent

Marking Agent

Fuzzy Inference Level

Collaboration Agent

Marking Agent

Marking Agent

Marking Agent

DiaryDiary

Diary

GroupingDegree

DisjoiningDegree

LastSituation

Fuzzy inference process

The Simulator

Agents representation

Parameters

Run Options

Testing Scenarios Scenario S1: 7 users Scenario S2: 4 users Scenario S3: 3 users

L = 50, 80 time steps

δ = 0.5, α = 0.1 Td = Tm = T = 60 sec

Conclusions and Future Works

Thanks for your attention

Scenario S1 – 7 Users

Ontological View

Repast 2.0

Framework for creating agent-based simulations using the Java language

Repast Engine Context Projections Agents Data Layers

Java Package: cr.agents

Class Diagram

Diary: Deterministic Finite Automaton

Membership Functions

Membership Functions

Fuzzy Sets – Grouping

LOW-MED

MED-HIGH

Fuzzy Sets – Disjoining

LOW-HIGH

Java Package: cr.core

Class Diagram

Sequence Diagram

Java Package: cr.services

Diffuse Service

Merge Service

Java Package: cr.fuzzy

Class Diagram

Sequence Diagram