Weaving Linked Open Data into User Profiling on the Social Web

DelftUniversity ofTechnology

Weaving Linked Open Data into User Profiling on the Social Web

MultiA-Pro, Lyon, France, April 16th 2012

Fabian Abel, Claudia Hauff, Geert-Jan Houben, Ke TaoWeb Information Systems, TU Delft, the Netherlands

2Weaving Linked Open Data into User Profiling on the Social Web

PersonalizedRecommendations

Personalized Search Adaptive Systems

What we do: Science and Engineering for the Personal Web

Social Web

Analysis and User Modeling

user/usage data

Semantic Enrichment, Linkage and Alignment

domains: news social media cultural heritage public data e-learning

recommending points of interest

3Weaving Linked Open Data into User Profiling on the Social Web

Kyle

Hometown: South Park, Colorado

4Weaving Linked Open Data into User Profiling on the Social Web

Kyle recently uploaded photos to Flickr

During his trip to the Netherlands, he uploaded pictures to Flickr.

tags: delft, vermeergeo: The Hague

tags: girl with pearl earringgeo: The Hague

5Weaving Linked Open Data into User Profiling on the Social Web

Kyle tweets about his upcoming trip

Looking forward to visit Paris next week!

6Weaving Linked Open Data into User Profiling on the Social Web

Interests of Kyle?

tags: delft, vermeergeo: The Hague

tags: girl with pearl earringgeo: The Hague

Looking forward to visit Paris next week!

• Given Kyle’s Flickr and Twitter activities, can we infer Kyle’s interests?

• Knowing that Kyle will visit Paris, France, can we recommend him places that might be interesting for him?

7Weaving Linked Open Data into User Profiling on the Social Web

Challenges• How to create a meaningful profile that supports

the given application? how to bridge between the Social Web chatter of a user and the candidate items of a recommender system?

User Profileconcept weight

?Application

that demands user interest profile regarding

-concepts

c1c5

c6

c9

Social Web d5

c1c3

c2

d1 d2d3 d4

d4 d3

Candidate Items Recommendations

8Weaving Linked Open Data into User Profiling on the Social Web

Challenge of Recommending Points of Interests (POIs) to Kyle

Johannes Vermeer

dbpedia:Louvre

Looking forward to visit Paris next week! dbpedia:Pari

s

The lacemaker

The astronomer

9Weaving Linked Open Data into User Profiling on the Social Web

c1

c4

c5

c6

LOD-based User Modelingweighting strategies

Applicationthat demands user

interest profile regarding -concepts

c2

c3

cx

cy

c9

User Profileconcept weight

0.4

0.1

0.2

c1

c2

c3

……

concepts that can be extracted from the user data

user data

Social Web

background knowledge (graph structures)

Linked Data

10Weaving Linked Open Data into User Profiling on the Social Web

User Modeling Building Blocks

11Weaving Linked Open Data into User Profiling on the Social Web

tags: girl with pearl earringgeo: The Hague

dbpedia:Girl_with_pearl_earring

A

ArtifactB

The lacemaker

C

Theastronomer

…

rdf:type

Johannes Vermeerfoaf:maker

foaf:maker

Strategies for exploiting the RDF-based background knowledge graph

Direct MentionIndirect Mention@RDF_statementIndirect Mention

@RDF_graph

dbpedia:Paris

dbpedia:Louvre dbpprop:locationlocatedIn

13Weaving Linked Open Data into User Profiling on the Social Web

Weighting Scheme

Weighting scheme: count the number of occurrences of a given graph pattern.

User Profileconcept weight

?

?

?

c1

c2

c3

……

User Profileconcept weight

374

152

73

c1

c2

c3

……

14Weaving Linked Open Data into User Profiling on the Social Web

Source of User Data• Twitter

• Flickr

15Weaving Linked Open Data into User Profiling on the Social Web

Mining the Geographic Origins of User Data

• Tweets or Flickr images posted by a GPS-enabled device;

• Images geo-tagged manually on Flickr world map

• Otherwise : exploit the title and the tags the users assign to their images

16Weaving Linked Open Data into User Profiling on the Social Web

Evaluation

17Weaving Linked Open Data into User Profiling on the Social Web

Research Questions

1. How does the source of user data influence the quality in deducing user preferences for POIs?

2. How does the consideration of background knowledge from the Linked Open Data Cloud impact the quality of the user modeling?

3. What (combination of) user modeling strategies allows for the best quality?

18Weaving Linked Open Data into User Profiling on the Social Web

Dataset

users 394

tweets 2,489,088 location 11% pictures 833,441 location 70.6%(within

10km)

duration 11 months

19Weaving Linked Open Data into User Profiling on the Social Web

Dataset Characteristics: Profile Sizes

81.4% of the users, Twitter-based profiles are

bigger than Flickr-based profiles

20Weaving Linked Open Data into User Profiling on the Social Web

Experimental Setup• Task:

= Recommending POIs = Predicting POIs which a user will visit

• Ground truth:• split data into training data (= first nine months) and test data

(= last two months) • POIs that the user visited in the last two months are considered

as relevant• Metrics:

• Precision@k, Recall@k and F-Measure@k: precision, recall and f-measure within the top k of the ranking of recommended items

21Weaving Linked Open Data into User Profiling on the Social Web

Results: Impact of User Data Source Selection

Combination of Twitter and Flickr

user data allows for best performance

Twitter seems to be more valuable for the

given application

22Weaving Linked Open Data into User Profiling on the Social Web

Results: Impact of Strategies for Exploiting RDF-based Background Knowledge

The more background information the better the user

modeling performance

23Weaving Linked Open Data into User Profiling on the Social Web

Results: Combining different Strategies

Combining all background exploitation strategies improves the user modeling performance

clearly

24Weaving Linked Open Data into User Profiling on the Social Web

ConclusionsWhat we did: • LOD-based User Modeling on the Social Web• Different strategies for exploiting RDF-based background

knowledgeFindings:1. Combination of different user data sources (Flickr &

Twitter) is beneficial for the user modeling performance2. User modeling quality increases the more background

knowledge one considers3. Combination of strategies achieves the best

performanceFuture work: •Investigate weighting schemes that weight the different

RDF graph patterns for acquiring background knowledge differently

25Weaving Linked Open Data into User Profiling on the Social Web

Thank you!

Slides : http://goo.gl/Zdg4K

Email: [email protected]: @wisdelft @taubauhttp://persweb.org