artefact-actor-networks at abis 2010
TRANSCRIPT
Modeling, obtaining and storing data from social media tools in Artefact-Actor-NetworksWolfgang Reinhardt, Tobias Varlemann, Matthias Moi, Adrian Wilke
University of Paderborn (Germany)Computer Science Education Group
Wolfgang Reinhardt, @wollepb, University of Paderborn (Germany)
Artefact-Actor-Networks
combination of
Social Networks
Artefact Networks
Social Media
Documents
goal
raise awareness about relevant people, topics and objects
Wolfgang Reinhardt, @wollepb, University of Paderborn (Germany)
Network of documentsNetwork in World Wide Web Consolidated artefact network I
Website A
Website B
Document C
Document D
(1) (2) (3)
D
CA
B
Consolidated artefact network IINetwork with bookmarksConsolidated artefact network I
Website B
Bookmark E
(1) (2) (3)
Wolfgang Reinhardt, @wollepb, University of Paderborn (Germany)
Consolidated actor networkActor network of company Private actor network
Person X
Person Y
Person X
Person Z
Person Z
Person X
Person Y
(1) (2) (3)
Consolidated artefact network II Consolidated actor network
Wolfgang Reinhardt, @wollepb, University of Paderborn (Germany)
Obtaining data
Wolfgang Reinhardt, @wollepb, University of Paderborn (Germany)
strong focus on Social Media tools
Delicious
Scribd
SlideShare
Blogs
Wikipedia
Scientific paper
Upload, DBLP, CiteSeer
Where does the data comes from?
Wolfgang Reinhardt, @wollepb, University of Paderborn (Germany)
How do we extract data?
Java-based backend
OSGi-enabled
hot deployment
Jena framework
crawl, store, analyse
crawl and parse
store data
analyse data
<< component >>Crawling-Block
<< component >>DataStore-Block
<< component >>Analyser-Block
Wolfgang Reinhardt, @wollepb, University of Paderborn (Germany)
Storing data
Wolfgang Reinhardt, @wollepb, University of Paderborn (Germany)
Wiki
WikiArtefact
WikiActor
DATA
MediaWiki
MediaWikiCategory
MediaWikiArtefact MediaWikiActor
DATA
Web
Webartefact
DATADATA
AANBase
Keyword
Artefact
Actor
DATA
hasMediaWikiCategory:hasKeyword
editedArticle:hasArtefact
pageIDoid
userComment
previousVersion:isRelatednextVersion:isRelated
screenName
linksTo:isRelatedhasPart:isRelated
redirectedFrom
creationTime
knows
isRelated hasArtefact
hasKeyword
KeywordValue
Dublin Core
SIOCFOAF
SWRC
Wolfgang Reinhardt, @wollepb, University of Paderborn (Germany)
ONTOLOGIES
Semantic relations
ART2 relations
between artefacts
isReplyOf, linksTo, hasPart, isReplyTo, hasComment
ACT2 relations
between actors
isFriendOf, relatesTo, collaboratesWith,
AA relations
between artefacts and actors
creatorOf, contributorOf, discussantOf, forwarderOf, bookmarkerOf
Wolfgang Reinhardt, @wollepb, University of Paderborn (Germany)
Analyzing data
Wolfgang Reinhardt, @wollepb, University of Paderborn (Germany)
Used analyzers
Text analyzers
Orchestr8 - Alchemy API
OpenCalais
Semantic similarity
SemSim algorithm
TF-IDF
Cosine similarity
Wolfgang Reinhardt, @wollepb, University of Paderborn (Germany)
Applicationsthat make use of Artefact-Actor-Networks
Wolfgang Reinhardt, @wollepb, University of Paderborn (Germany)
Wolfgang Reinhardt, @wollepb, University of Paderborn (Germany)
0.73
0.73
0.73
0.15
0.36
0.15
0.38
3.04
0.11
0.17
0.42
1.31
0.91
0.56
0.28
0.5
0.29
0.66
7.0
0.78
1.61
0.11
0.16
0.75
0.11
0.56
0.28
0.5
0.29
0.66
7.0
1.44
0.17
0.28
0.28
0.24
0.35
0.45
5.29
0.17
0.73
0.32
2.77
0.73
0.5
0.5
0.24
0.25
0.8
8.87
0.42
0.16
0.63
0.42
0.151.05
0.71
0.97
2.53
0.69
0.36
0.29
0.29
0.35
0.25
1.05
1.05
1.67
0.46
9.3
0.15
0.71
1.05
0.97
0.34
0.38
0.32
0.97
1.67
0.97
0.7
0.66
0.78
0.66
0.45
0.8
0.46
10.78
0.73
3.04
1.31
7.0
1.61
0.75
7.0
1.44
5.29
2.77
8.87
0.63
2.53
9.3
0.34
0.7
10.78
1.31
0.73
0.91
0.11
0.17
0.42
1.31
0.73
0.73
0.73
0.69
0.5263157894736842
0.5714285714285714
0.5555555555555556
0.5714285714285714
0.5128205128205129
0.5405405405405405
0.5714285714285714
0.5
0.6060606060606061
0.5405405405405405
0.5714285714285714
0.5405405405405405
0.5882352941176471
0.6666666666666666
0.5714285714285714
0.5882352941176471
0.5882352941176471
0.9523809523809523
0.5555555555555556
0.5263157894736842
0.37735849056603776
Wolfgang Reinhardt, @wollepb, University of Paderborn (Germany)
Analysis of students’ Social Media use
Wolfgang Reinhardt, @wollepb, University of Paderborn (Germany)
Wolfgang Reinhardt, @wollepb, University of Paderborn (Germany)
Wolfgang Reinhardt, @wollepb, University of Paderborn (Germany)
What’s next?
deal with performance issues
1.5M Wikipedia articles & >> 4B RDF triples
Jena & SPARQL = no good
Reasoning & Inferencing of large data sets = ouch
Recommender Systems
Clustering
Advanced semantic similarity
Wolfgang Reinhardt, @wollepb, University of Paderborn (Germany)
Thank you for your attention
Wolfgang Reinhardt, @wollepbUniversity of Paderborn, Germany
Wolfgang Reinhardt, @wollepb, University of Paderborn (Germany)