BOTTARI: How to offer innovative services listening to the Social Media

BOTTARI: How to offer innovative services

listening to the Social Media

Emanuele Della Valle DEI - Politecnico di Milano

[email protected]://emanueledellavalle.org

• Introduction

• Architecture

• Components

• Query Processing

• Evaluation– Efficacy– Efficiency

• Conclusions

Agenda

227.02.2012 - Workshop di Senso Comune, Trento, Italy


Introduction

A stream of live opinions

• Micro-posts carry the live opinion of social media users about topics and points of interest.


• In 2011, an average of three million tweets per day was posted in Seoul.

• A large number of these micro-posts are about restaurants, bars, coffees and many other semi-public points of interest (POIs) in the city.

• Let us consider, Insadong district with its +200 restaurants

Introduction

Seoul’s stream of live opinions


Number of tweets per day related to Insadong’srestaurants

Insadong’s restaurantsordered by decreasingfrequency of tweets

• Trusting this collective opinion, can novel commercial and social services be considered?

• The case of BOTTARI

Introduction

A commercial/social opportunity?


http://www.youtube.com/watch?v=XGOKe_lhSks

• An augmented reality application for personalized recommendation of restaurants in Insadong (Seoul)

Introduction

What have you seen?


• Yes and no!

• Same use case

• BOTTARI listens to the social media

Introduction

Yet another ?


outout

Query RewriterQuery Rewriter

Query Evaluator

Query Evaluator

Machine LearningMachine Learning

Streaming ProcessorStreaming Processor

Geo Filtering

Geo Filtering

Geo-spatial KB

Geo-spatial KB

Social Media Analyzer


Architecture

Android Client

27.02.2012 - Workshop di Senso Comune, Trento, Italy 8

outout


Query Evaluator

Query Evaluator


Semantic Stream

Processor

Semantic Stream

Processor

Geo Filtering

Geo Filtering

Geo-spatial KB



Architecture

Geo-Spatial KB


outout


Query Evaluator

Query Evaluator


Semantic Stream

Processor

Geo Filtering

Geo Filtering

Geo-spatial KB

Geo-spatial KB


PUSH: Data Initiated

ArchitecturePUSH Segment – Data Initiated


out

Query Rewriter

Query Evaluator

Machine Learning

SemanticStream

Processor

SemanticStream

Processor

Geo Filtering

Geo-spatial KB

Geo-spatial KB



PU

LL

: Qu

ery

Initi

ate

d

ArchitecturePULL Segment – Query Initiated


out

Query Rewriter

Query Evaluator

Machine Learning

Semantic Stream

Processor

Geo Filtering

Geo-spatial KB


PU

LL

: Qu

ery

Initi

ate

d


ArchitectureThe Entire BOTTARI System


Components

SOR - Geo-Spatial KB


Components



Micro-post

MorphologicallyAnalyzable?

Rule based Analysis

Auto generated rules

Auto generated rules

Learneddocuments

SVMs

Syllable KernelSyllable Kernel

Sentiment of the tweet

Yes No

• Precision tests:– Auto-generated

rules ≈ 70%

– Manually-coded rules ≈ 90%

– Syllable kernel ≈ 50~60%

• Our target > 85%

Components

Semantic Stream Processor


• SUNS (Statistical Unit Node Sets)– A machine learning framework for inductive

materialization• Detects interesting data patterns• Predics RDF-triples

– i.e., which restaurant a user will tweet positively about

– Caractheristics • Capability to deal with sparse, high-dimensional and

incomplete data• Multivariate latent space based approach• Modularized approach for easily integrating contextual

information

Components

Machine Learning Component


out

Query Rewriter

Query Evaluator

Machine Learning

Semantic Stream

Processor

Geo Filtering

Geo-spatial KB


PU

LL

: Qu

ery

Initi

ate

d


Query ProcessingAn Ontology-based Approach


17

SELECT DISTINCT ?poi ?name ?lat ?long ?numPos ?prob WHERE { ?poi a ns:NamedPlace ; ns:name ?name ; geo:lat ?lat ; geo:long ?long . FILTER (f:within_distance(37.5, 126.9, ?lat, ?long, 200)) FILTER (f:dest_point_viewing(37.5, 126.9, ?lat, ?long, 90, 200)) { :someUser sioc:creator_of ?tweet . ?tweet twd:talksAboutPositively ?poi . WITH PROBABILITY ?prob ENSURE PROBABILITY [0.5..1) } ?poi twd:numberOfPositiveTweets ?numPos . } ORDER BY DESC(?numPos), ?prob, f:distance(37.5, 126.9, ?lat, ?long)LIMIT 10


GEO-SPATIAL FILTER

RECCOMENDENDATION

REPUTATION

Query Processing

An Example Query

outQuery Evaluator

Query Rewriter

Machine Learning

Semantic Stream

Processor

Geo Filtering

Geo-spatial KB


The issued query is rewritten in three queries; one for each

run time component

PU

LL

: Qu

ery

Initi

ate

d


Query ProcessingThe Input Received is Rewritten


outQuery Evaluator

Query Rewriter

Machine Learning

Semantic Stream

Processor

Geo Filtering

Geo-spatial KB


Generates personalized recommendations

PU

LL

: Qu

ery

Initi

ate

d


Query ProcessingEach Component Answers


outQuery Evaluator

HTTP

Query Rewriter

Machine Learning

Semantic Stream

Processor

Geo Filtering

Geo-spatial KB


Gets trends in restaurants’ social reputation P

UL

L: Q

ue

ry In

itiat

ed




outQuery Evaluator

Query Rewriter

Machine Learning

Semantic Stream

Processor

Geo Filtering

Geo-spatial KB


Gets restaurants

close to user location

PU

LL

: Qu

ery

Initi

ate

d




outQuery Evaluator

Joins results and returns

them

Query Rewriter

Machine Learning

Semantic Stream

Processor

Geo Filtering

Geo-spatial KB


PU

LL

: Qu

ery

Initi

ate

d


Query ProcessingResults Are Joined and Returned


23

Evaluation

Efficacy – High Accuracy


Evaluation

Efficacy – Remembering 90 Days is Enough


Evaluation

Efficiency – High Throughput


• End-user application

• Attractive and functional interface

• Real-world dynamic data

• Fully based on Semantic Web technologies– RDF as common data format between heterogeneous

components– SPARQL as query language

• Rigorously evaluated– Effective– High throughput for handling dynamic data

• Commercial Potential

Conclusions


• Tony Lee, Seonho Kim (SALTLUX)– Android Application – Social Media Analyzer– Tweets and restaurant descriptiond for Insadong

• Marco Balduini (PoliMi)– Semantic Stream Processor

• Irene Celino, Daniele Dell’Aglio (CEFRIEL)– LarKC workflows– Ontology

• Volker Tresp, Yi Huang (SIEMENS)– Machine Learning component

Credits

BOTTARI is a joint work with


Thank You! Questions?


Much More to Come!Keep an eye on

http://www.streamreasoning.org

BOTTARI: How to offer innovative services

listening to the Social Media

Emanuele Della Valle DEI - Politecnico di Milano

[email protected]://emanueledellavalle.org

BOTTARI: How to offer innovative services listening to the Social Media

Business

Transcript of BOTTARI: How to offer innovative services listening to the Social Media