Heiko Paulheim, profile picture
Uploaded byHeiko Paulheim
ODP, PPTX4,823 views

Exploiting Linked Open Data as Background Knowledge in Data Mining

The document summarizes an approach to exploiting linked open data as background knowledge in data mining tasks. It describes using LOD to generate additional features for machine learning algorithms from entity names in datasets. Experiments show this approach can improve results for classification tasks. Applications discussed include classifying events from Wikipedia and tweets by leveraging background knowledge from DBpedia to prevent overfitting. The document also proposes using LOD to help explain statistics by enriching datasets and analyzing correlations.

Embed presentation

Download as ODP, PPTX
10/08/13 Heiko Paulheim 1 Exploiting Linked Open Data as Background Knowledge in Data Mining Heiko Paulheim, University of Mannheim
10/08/13 Heiko Paulheim 2 Outline • Motivation • The original FeGeLOD framework • Experiments • Applications • The RapidMiner Linked Open Data Extension • Challenges and Future Work
10/08/13 Heiko Paulheim 3 Motivation: An Example Data Mining Task • Analyzing book sales ISBN City Sold 3-2347-3427-1 Darmstadt 124 3-43784-324-2 Mannheim 493 3-145-34587-0 Roßdorf 14 ... ISBN City Population ... Genre Publisher ... Sold 3-2347-3427-1 Darm- stadt 144402 ... Crime Bloody Books ... 124 3-43784-324-2 Mann- heim 291458 … Crime Guns Ltd. … 493 3-145-34587-0 Roß- dorf 12019 ... Travel Up&Away ... 14 ... → Crime novels sell better in larger cities
10/08/13 Heiko Paulheim 4 Motivation • Many data mining problems are solved better – when you have more background knowledge (leaving scalability aside) • Problems: – Tedious work – Selection bias: what to include?
10/08/13 Heiko Paulheim 5 Motivation http://lod-cloud.net/
10/08/13 Heiko Paulheim 6 Motivation • Idea: – reuse background knowledge from Linked Open Data – include it in the data mining process as needed • Two main variants: – develop mining/learning algorithms that run directly on Linked Data – create relational features from Linked Data
10/08/13 Heiko Paulheim 7 Motivation • Develop mining/learning algorithms – e.g., DL Learner – e.g., dedicated Kernel functions • Advantages: – can be quite efficient – no reduction to “flat” table structure – semantics can be respected directly
10/08/13 Heiko Paulheim 8 Motivation • Create relational features – e.g., LiDDM – e.g., AutoSPARQL – e.g., FeGeLOD / RapidMiner Linked Open Data Extension • Advantages: – Easy combination of knowledge from various sources • including relational features in the original data – Arbitrary mining algorithms/tools possible
10/08/13 Heiko Paulheim 9 FeGeLOD – Feature Generation from LOD IS B N 3 -2 3 4 7 -3 4 2 7 -1 C ity D a r m s ta d t # s o ld 1 2 4 N a m e d E n t it y R e c o g n it io n IS B N 3 -2 3 4 7 -3 4 2 7 - 1 C ity D a r m s ta d t # s o ld 1 2 4 C ity _ U R I h ttp : / / d b p e d ia .o r g / r e s o u r c e/ D a r m s ta d t F e a t u r e G e n e r a t io n IS B N 3 -2 3 4 7 -3 4 2 7 -1 C ity D a r m s ta d t # s o ld 1 2 4 C ity _ U R I h ttp : / / d b p e d ia .o r g / r e s o u r c e / D a r m s ta d t C ity _ U R I_ d b p e d ia -o w l: p o p u la tio n T o ta l 1 4 1 4 7 1 C ity _ U R I_ ... ... F e a t u r e S e le c t io n IS B N 3 -2 3 4 7 -3 4 2 7 - 1 C ity D a r m s ta d t # s o ld 1 2 4 C ity _ U R I h ttp : / / d b p e d ia .o r g / r e s o u r c e/ D a r m s ta d t C ity _ U R I_ d b p e d ia -o w l:p o p u la tio n T o ta l 1 4 1 4 7 1
10/08/13 Heiko Paulheim 10 FeGeLOD – Feature Generation from LOD • Original prototype, based on Weka: – Simple NER (guessing URIs) – Seven generators: • direct types • data properties • unqualified relations (boolean, numeric) • qualified relations (boolean, numeric) • individuals (dangerous!) - may be restricted to specific property – Simple feature selection: filtering features • that have only* different values (expect numerical) • that have only* identical values • that are mostly missing* *) 95% or 99%
10/08/13 Heiko Paulheim 11 Experiments • Testing with two* standard machine learning data sets – Zoo: classifying animals – AAUP: predicting income of university employees (regression task) • Question: how much improvement do additional features bring? *) standard ML datasets with speaking labels are scarce!
10/08/13 Heiko Paulheim 12 Experiments: Zoo Dataset
10/08/13 Heiko Paulheim 13 First Results: AAUP
10/08/13 Heiko Paulheim 14 Experiments: Early Insights • Additional features often improve the results • Zoo dataset: – Ripper: 89.11 to 96.04 – SMO: 93.07 to 97.03 – No improvement for Naive Bayes • AAUP dataset (compensation): – M5: 59.88 to 51.28 – SMO: 74.12 to 61.97 – No improvement for linear regression • ...but they may also cause problems – extreme example: 6.54 to 189.90 for linear regression – memory and timeouts due to large datasets
10/08/13 Heiko Paulheim 15 Experiments: Quality of Features • Information gain of features on Zoo dataset
10/08/13 Heiko Paulheim 16 Experiments: Quality of Features • Information gain of features on AAUP dataset (compensation)
10/08/13 Heiko Paulheim 17 Application: Classifying Events from Wikipedia • Event Extraction from Wikipedia • Joint work with Dennis Wegener and Daniel Hienert (GESIS) • Task: event classification (e.g., Politics, Sports, ...) http://www.vizgr.org/historical-events/timeline/
10/08/13 Heiko Paulheim 18 Application: Classifying Events from Wikipedia • Source Material: http://www.vizgr.org/historical-events/timeline/
10/08/13 Heiko Paulheim 19 Application: Classifying Events from Wikipedia • Positive Examples for class politics: – 2011, March 15 - German chancellor Angela Merkel shuts down the seven oldest German nuclear power plants. – 2010, June 3 – Christian Wulff is nominated for President of Germany by Angela Merkel. • Negative Examples for class politics: – 2010, July 7 – Spain defeats Germany 1-0 to win its semi-final and for its first time, along with Netherlands make the 2010 FIFA World Cup Final. – 2012, February 16 – Roman Lob is selected to represent Germany in the Eurovision Song Contest.
10/08/13 Heiko Paulheim 20 Application: Classifying Events from Wikipedia • Positive Examples for class politics: – 2011, March 15 - German chancellor Angela Merkel shuts down the seven oldest German nuclear power plants. – 2010, June 3 – Christian Wulff is nominated for President of Germany by Angela Merkel. • Negative Examples for class politics: – 2010, July 7 – Spain defeats Germany 1-0 to win its semi-final and for its first time, along with Netherlands make the 2010 FIFA World Cup Final. – 2012, February 16 – Roman Lob is selected to represent Germany in the Eurovision Song Contest. • Possible learned model: – "Angela Merkel" → Politics
10/08/13 Heiko Paulheim 21 Application: Classifying Events from Wikipedia • Possibly Learned Model: – "Angela Merkel" → Politics • How can we do better? • Background knowledge from Linked Open Data – 2011, March 15 - German chancellor Angela Merkel [class: Politician] shuts down the seven oldest German nuclear power plants. – 2012, May 13, Elections in North Rhine-Westphalia – Hannelore Kraft [class: Politician] is elected to continue as Minister-President, heading an SPD- Green coalition. • Model learned in that case: – "[class: Politician]" → Politics
10/08/13 Heiko Paulheim 22 Application: Classifying Events from Wikipedia • Model learned in that case: – "[class: Politician]" → Politics • Much more general – Can also classify events with politicians not contained in the training set • Less training examples required – A few events with politicians, athletes, singers, ... are enough
10/08/13 Heiko Paulheim 23 Application: Classifying Events from Wikipedia • Experiments on Wikipedia data – >10 categories – 1,000 labeled examples as training set – Classification accuracy: 80% • Plus: – We have trained a language-independent model! • often, models are like "elect*" → Politics – 22. Mai 2012: Peter Altmaier [class: Politician] wird als Nachfolger von Norbert Röttgen [class: Politician] zum Bundesumweltminister ernannt. – 6 januari 2012: Jonas Sjöstedt [class: Politician] väljs till ny partiledare för Vänsterpartiet efter Lars Ohly [class: Politician].
10/08/13 Heiko Paulheim 24 Application: Classifying Tweets • Joint work with Axel Schulz and Petar Ristoski (SAP Research) • Goal: using Twitter for emergency management fire at #mannheim #universityomg two cars on fire #A5 #accident fire at train station still burning my heart is on fire!!!come on baby light my fire boss should fire that stupid moron
10/08/13 Heiko Paulheim 25 Application: Classifying Tweets • Social media contains data on many incidents – But keyword search is not enough – Detecting small incidents is hard – Manual inspection is too expensive (and slow) • Machine learning could help – Train a model to classify incident/non incident tweets – Apply model for detecting incident related tweets • Training data: – Traffic accidents – ~2,000 tweets containing relevant keywords (“car”, “crash”, etc.), hand labeled (50% related to traffic incidents)
10/08/13 Heiko Paulheim 26 Application: Classifying Tweets • Learning to classify tweets: – Positive and negative examples – Features: • Stemming • POS tagging • Word n-grams • … • Accuracy ~90% • But – Accuracy drops to ~85% when applying the model to a different city
10/08/13 Heiko Paulheim 27 Application: Classifying Tweets • Example set: – “Again crash on I90” – “Accident on I90” • Model: – “I90” → indicates traffic accident • Applying the model: – “Two cars crashed on I51” → not related to traffic accident
10/08/13 Heiko Paulheim 28 Using LOD for Preventing Overfitting • Example set: – “Again crash on I90” – “Accident on I90” dbpedia:Interstate_90 dbpedia-owl:Road rdf:type dbpedia:Interstate_51 rdf:type • Model: – dbpedia-owl:Road → indicates traffic accident • Applying the model: – “Two cars crashed on I51” → indicates traffic accident • Using DBpedia Spotlight + FeGeLOD – Accuracy keeps up at 90% – Overfitting is avoided
10/08/13 Heiko Paulheim 29 Explaining Statistics • Statistics are very wide spread – Quality of living in cities – Corruption by country – Fertility rate by country – Suicide rate by country – Box office revenue of films – ...
10/08/13 Heiko Paulheim 30 Explaining Statistics • Questions we are often interested in – Why does city X have a high/low quality of living? – Why is the corruption higher in country A than in country B? – Will a new film create a high/low box office revenue? • i.e., we are looking for – explanations – forecasts (e.g., extrapolations)
10/08/13 Heiko Paulheim 31 Explaining Statistics http://xkcd.com/605/
10/08/13 Heiko Paulheim 32 Explaining Statistics • What statistics often look like
10/08/13 Heiko Paulheim 33 Explaining Statistics • There are powerful tools for finding correlations etc. – but many statistics cannot be interpreted directly – background knowledge is missing • Approach: – use Linked Open Data for enriching statistical data (e.g., FeGeLOD) – run analysis tools for finding explanations
10/08/13 Heiko Paulheim 34 Prototype Tool: Explain-a-LOD • Loads a statistics file (e.g., CSV) • Adds background knowledge • Runs basic analysis (correlation, rule learning) • Presents explanations
10/08/13 Heiko Paulheim 35 Statistical Data: Examples • Data Set: Mercer Quality of Living – Quality of living in 216 cities word wide – norm: NYC=100 (value range 23-109) – As of 1999 – http://across.co.nz/qualityofliving.htm • LOD data sets used in the examples: – DBpedia – CIA World Factbook for statistics by country
10/08/13 Heiko Paulheim 36 Statistical Data: Examples • Examples for low quality cities – big hot cities (junHighC >= 27 and areaTotalKm >= 334) – cold cities where no music has ever been recorded (recordedIn_in = false and janHighC <= 16) – latitude <= 24 and longitude <= 47 • a very accurate rule • but what's the interpretation? Next Record Studio 2547 miles Next Record Studio 2547 miles
10/08/13 Heiko Paulheim 37 Statistical Data: Examples
10/08/13 Heiko Paulheim 38 Statistical Data: Examples • Data Set: Transparency International – 177 Countries and a corruption perception indicator (between 1 and 10) – As of 2010 – http://www.transparency.org/cpi2010/results
10/08/13 Heiko Paulheim 39 Statistical Data: Examples • Example rules for countries with low corruption – HDI > 78% • Human Development Index, calculated from live expectancy, education level, economic performance – OECD member states – Foundation place of more than nine organizations – More than ten mountains – More than ten companies with their headquarter in that state, but less than two cargo airlines
10/08/13 Heiko Paulheim 40 Statistical Data: Examples • Data Set: Burnout rates – 16 German DAX companies – Absolute and relative numbers – As of 2011 – http://de.statista.com/statistik/daten/studie/226959/umfrage/burn-out- erkrankungen-unter-mitarbeitern-ausgewaehlter-dax-unternehmen/
10/08/13 Heiko Paulheim 41 Datavalues Type Unqualifiedrelation(boolean) Unqualifiedrelation(numeric) Qualifiedrelation(boolean) Qualifiedrelation(numeric) Joint 1 1.5 2 2.5 3 3.5 4 4.5 5 Correlation Rule Learning Evaluation of Feature Quality • Quality of living dataset
10/08/13 Heiko Paulheim 42 Datavalues Type Unqualifiedrelation(boolean) Unqualifiedrelation(numeric) Qualifiedrelation(boolean) Qualifiedrelation(numeric) Joint 1 1.5 2 2.5 3 3.5 4 4.5 5 Correlation Rule Learning Evaluation of Feature Quality • Corruption dataset
10/08/13 Heiko Paulheim 43 Statistical Data: Examples • Findings for burnout rates – Positive correlation between turnover and burnout rates – Car manufacturers are less prone to burnout – German companies are less prone to burnout than international ones • Exception: Frankfurt
10/08/13 Heiko Paulheim 44 Statistical Data: Examples • Data Set: Antidepressives consumption – In European countries – Source: OECD – http://www.oecd-ilibrary.org/social-issues-migration-health/health-at-a-glance- 2011/pharmaceutical-consumption_health_glance-2011-39-en
10/08/13 Heiko Paulheim 45 Statistical Data: Examples • Findings for antidepressives consumption – Larger countries have higher consumption – Low HDI → high consumption – By geography: • Nordic countries, countries at the Atlantic: high • Mediterranean: medium • Alpine countries: low – High average age → high consumption – High birth rates → high consumption
10/08/13 Heiko Paulheim 46 Statistical Data: Examples • Data Set: Suicide rates – By country – OECD states – As of 2005 – http://www.washingtonpost.com/wp-srv/world/suiciderate.html
10/08/13 Heiko Paulheim 47 Statistical Data: Examples • Findings for suicide rates – Democraties have lower suicide rates than other forms of government – High HDI → low suicide rate – High population density → high suicide rate – By geography: • At the sea → low • In the mountains → high – High Gini index → low suicide rate • High Gini index ↔ unequal distribution of wealth – High usage of nuclear power → high suicide rates
10/08/13 Heiko Paulheim 48 Statistical Data: Examples • Data set: sexual activity – Percentage of people having sex weekly – By country – Survey by Durex 2005-2009 – http://chartsbin.com/view/uya
10/08/13 Heiko Paulheim 49 Statistical Data: Examples • Findings on sexual activity – By geography: • High in Europe, low in Asia • Low in Island states – By language: • English speaking: low • French speaking: high – Low average age → high activity – High GDP per capita → low activity – High unemployment rate → high activity – High number of ISP providers → low activity
10/08/13 Heiko Paulheim 50 Try it... but be careful! • Download from http://www.ke.tu-darmstadt.de/resources/explain-a-lod • including a demo video, papers, etc. http://xkcd.com/552/
10/08/13 Heiko Paulheim 51 RapidMiner Linked Open Data Extension • August 16th , 2013: FeGeLOD celebrates its 2nd birthday • Problems – still no nice UI – special configurations are tricky – difficult to enhance • Decision – Reimplementation on RapidMiner platform – September 13th , 2013: Release of RapidMiner Linked Open Data Extension – Available from RapidMiner marketplace • http://dws.informatik.uni-mannheim.de/en/research/rapidminer-lod-extension/
10/08/13 Heiko Paulheim 52 RapidMiner Linked Open Data Extension • Simple wiring of operators – linkers – generators • Combination with powerful RapidMiner operators
10/08/13 Heiko Paulheim 53 RapidMiner Linked Open Data Extension • Easy SPARQL endpoint definitions • Support of custom SPARQL statements
10/08/13 Heiko Paulheim 54 Challenges and Future Work • SPARQL variants – Some endpoints support special/non-standard SPARQL constructs – COUNT(...) – transitive closure – exploit where applicable • Implementations without SPARQL – Freebase – OpenCyc
10/08/13 Heiko Paulheim 55 Challenges and Future Work • Linking is still challenging – URI patterns are not flexible – Search by label is time consuming – Services like DBpedia Lookup are scarce • Limitations of completely unsupervised linking – e.g., Hurricanes – how to use headlines/attribute names?
10/08/13 Heiko Paulheim 56 Challenges and Future Work • Linking as optimization problem – find candidates for all entities, e.g., by DBpedia lookup – find a selection of candidates that are most similar to each other • e.g., all of them are U.S. cities – some experiments with types and categories • problem: not complete – some problems cannot be addressed (e.g.: Hurricanes) • Alternatives: – semi supervised linking – user provides some example links – active learning
10/08/13 Heiko Paulheim 57 Challenges and Future Work • Exploiting semantics for feature selection • Given two features: – f1: type(RoadsInAlaska) – f2: type(Road) • and the schema definition Road rdfs:subclassOf RoadsInAlaska • Exploit that information for feature selection – e.g., gain(f1) ≈ gain(f2), f1<f2 → remove f1
10/08/13 Heiko Paulheim 58 Challenges and Future Work • Incompleteness of LOD – e.g., type information in DBpedia – may lead to findings such as • if a city is of type Place, the quality of living is high – possible remedy: autocomplete on the dataset (e.g., Paulheim/Bizer 2013) • Biases in LOD – e.g., DBpedia has a bias towards western culture – may lead to findings such as • if many records have been made in a city, the quality of living is high
10/08/13 Heiko Paulheim 59 Challenges and Future Work • Features not used for scalability reasons: – features for single entities • e.g., “Roman Polanski directorOf X” – features more than one hop away • e.g., “Cities with a university which has a computer science department” – some are covered by YAGO types, e.g., “AustralianBandsFoundedIn1990” • but subject to YAGO's selection bias • Approaches are required to use such features – which respect scalability – “generate first, filter later” is not the best solution • e.g., “Cities with at least one of ArtSchoolsInParis” – on-the-fly filtering may be more suitable • e.g., sampling
10/08/13 Heiko Paulheim 60 Challenges and Future Work • Automatically exploit data sources with non-simple structures EU18931 a Funding . EU18931 has-grant-value [ has-amount 1300000 . has-unit-of-measure EUR . ] • Support geo/temporal features – e.g., Data Cubes – e.g., Linked Geo Data • Construct complex features (in a scalable way!) – e.g., cinemas per inhabitant real example from CORDIS dataset
10/08/13 Heiko Paulheim 61 Wrap-up • Linked Data is useful as background knowledge – especially on problems which have little knowledge in themselves • Unsupervised methods – avoid biases and work without knowledge about LOD – but: scalability and generality problems • RapidMiner LOD extension – a constantly growing toolkit
10/08/13 Heiko Paulheim 62 Credits & Thanks • Past contributors of FeGeLOD: – Johannes Fürnkranz – Raad Bahmani – Alexander Gabriel – Simon Holthausen • Current team of RapidMiner Linked Open Data Extension: – Chris Bizer – Petar Ristoski – Evgeny Mitichkin
10/08/13 Heiko Paulheim 63 Exploiting Linked Open Data as Background Knowledge in Data Mining Heiko Paulheim, University of Mannheim

More Related Content

ODP
Data Mining with Background Knowledge from the Web - Introducing the RapidMin...
byHeiko Paulheim
 
PPT
Big Data, Smart Algorithms, and Market Power - A Computer Scientist’s Perspec...
byHeiko Paulheim
 
ODP
Big Data, Smart Algorithms, and Market Power - A Computer Scientist's Perspec...
byHeiko Paulheim
 
ODP
Machine Learning & Embeddings for Large Knowledge Graphs
byHeiko Paulheim
 
PDF
Ld4 dh tutorial
byEnrico Daga
 
ODP
Mining the Web of Linked Data with RapidMiner
byHeiko Paulheim
 
PPTX
Immigration Reporting: Data, Tools and Sources
byborderzine
 
PDF
New Adventures in RDF2vec
byHeiko Paulheim
 
Data Mining with Background Knowledge from the Web - Introducing the RapidMin...
byHeiko Paulheim
 
Big Data, Smart Algorithms, and Market Power - A Computer Scientist’s Perspec...
byHeiko Paulheim
 
Big Data, Smart Algorithms, and Market Power - A Computer Scientist's Perspec...
byHeiko Paulheim
 
Machine Learning & Embeddings for Large Knowledge Graphs
byHeiko Paulheim
 
Ld4 dh tutorial
byEnrico Daga
 
Mining the Web of Linked Data with RapidMiner
byHeiko Paulheim
 
Immigration Reporting: Data, Tools and Sources
byborderzine
 
New Adventures in RDF2vec
byHeiko Paulheim
 

Viewers also liked

PDF
Proposal for a quality framework for the evaluation of administrative and sur...
byPiet J.H. Daas
 
DOCX
Quality management system procedures
byselinasimpson2101
 
PPTX
Quality framework
bysaurabhshri
 
PDF
Bpo risk management 2013
byRahul Bhan (CA, CIA, MBA)
 
PDF
WebeX Presentation - Quality Consortium
byThe Avoca Group
 
DOCX
Sharepoint quality management system
byselinasimpson2101
 
PDF
Mixed Methods Research
byRoller Research
 
PDF
Process asset library as process improvement and knowledge sharing tool
byKobi Vider
 
PDF
2004 E2M - The ShopView Story Information Package.PDF
byMelissa Jones
 
DOCX
QMS SharePoint Structure Definition Document
byMelissa Jones
 
DOCX
Part 3 - SharePoint QMS Anyone Can Make - Data Dictionary
byMelissa Jones
 
DOCX
QMS SharePoint Wireframe - download and edit for you use
byMelissa Jones
 
PPTX
Quality framework 1
byShwetha Bhat
 
PPTX
Metadata Quality Assurance Framework at QQML2016 conference - full version
byPéter Király
 
PPTX
Quality measurement - How to measure the quality of any object?
byGrzegorz Grela
 
PPTX
Audit Quality Framework & Proportionate Application of ISAs
byInternational Federation of Accountants
 
PPT
15 Months to Certification: Using SharePoint as the Platform for an ISO 9001 ...
byBarry Peters
 
PPTX
QMS Calibration Powerpoint
byDennis J Morgan
 
PPTX
PAS: The Planning Quality Framework
byPAS_Team
 
PPTX
Solr 6.0 Graph Query Overview
byKevin Watters
 
Proposal for a quality framework for the evaluation of administrative and sur...
byPiet J.H. Daas
 
Quality management system procedures
byselinasimpson2101
 
Quality framework
bysaurabhshri
 
Bpo risk management 2013
byRahul Bhan (CA, CIA, MBA)
 
WebeX Presentation - Quality Consortium
byThe Avoca Group
 
Sharepoint quality management system
byselinasimpson2101
 
Mixed Methods Research
byRoller Research
 
Process asset library as process improvement and knowledge sharing tool
byKobi Vider
 
2004 E2M - The ShopView Story Information Package.PDF
byMelissa Jones
 
QMS SharePoint Structure Definition Document
byMelissa Jones
 
Part 3 - SharePoint QMS Anyone Can Make - Data Dictionary
byMelissa Jones
 
QMS SharePoint Wireframe - download and edit for you use
byMelissa Jones
 
Quality framework 1
byShwetha Bhat
 
Metadata Quality Assurance Framework at QQML2016 conference - full version
byPéter Király
 
Quality measurement - How to measure the quality of any object?
byGrzegorz Grela
 
Audit Quality Framework & Proportionate Application of ISAs
byInternational Federation of Accountants
 
15 Months to Certification: Using SharePoint as the Platform for an ISO 9001 ...
byBarry Peters
 
QMS Calibration Powerpoint
byDennis J Morgan
 
PAS: The Planning Quality Framework
byPAS_Team
 
Solr 6.0 Graph Query Overview
byKevin Watters
 

Similar to Exploiting Linked Open Data as Background Knowledge in Data Mining

ODP
Linked Open Data enhanced Knowledge Discovery
byHeiko Paulheim
 
PDF
Text Classification Powered by Apache Mahout and Lucene
bylucenerevolution
 
PPT
Unexpected Challenges in Large Scale Machine Learning by Charles Parker
byBigMine
 
PPT
What happen after crawling big data?
byjcarpioc
 
PPTX
A Comparison of Propositionalization Strategies for Creating Features from Li...
byPetar Ristoski
 
PDF
A Comparative Study of different Classifiers on Political Data for Classifica...
byIRJET Journal
 
PPTX
Chapter 2 Introduction to CR_Process.pptx
byTitiA3
 
PDF
Cedem futia-2014
byDanube University Krems, Centre for E-Governance
 
PPTX
Language of Politics on Twitter - 03 Analysis
byYelena Mejova
 
PDF
Machine Learning: Learning with data
byONE Talks
 
PDF
One talk Machine Learning
byONE Talks
 
PDF
Analyzing-Threat-Levels-of-Extremists-using-Tweets
byRESHAN FARAZ
 
PPT
Data Science in the Real World: Making a Difference
bySrinath Perera
 
PPTX
LOD2 Webinar Series FOX
byLOD2 Creating Knowledge out of Interlinked Data
 
PPT
The science behind predictive analytics a text mining perspective
byankurpandeyinfo
 
PPT
CityPulse - Wright State University
byCityPulse Project
 
PDF
A Quick Overview of Artificial Intelligence and Machine Learning
byHiroki Sayama
 
PPT
ppt
bybutest
 
PDF
Data: Past, Present, and Future (Lecture 1, Spring 2018)
bychris wiggins
 
PDF
ESWC SS 2012 - Friday Keynote Marko Grobelnik: Big Data Tutorial
byeswcsummerschool
 
Linked Open Data enhanced Knowledge Discovery
byHeiko Paulheim
 
Text Classification Powered by Apache Mahout and Lucene
bylucenerevolution
 
Unexpected Challenges in Large Scale Machine Learning by Charles Parker
byBigMine
 
What happen after crawling big data?
byjcarpioc
 
A Comparison of Propositionalization Strategies for Creating Features from Li...
byPetar Ristoski
 
A Comparative Study of different Classifiers on Political Data for Classifica...
byIRJET Journal
 
Chapter 2 Introduction to CR_Process.pptx
byTitiA3
 
Cedem futia-2014
byDanube University Krems, Centre for E-Governance
 
Language of Politics on Twitter - 03 Analysis
byYelena Mejova
 
Machine Learning: Learning with data
byONE Talks
 
One talk Machine Learning
byONE Talks
 
Analyzing-Threat-Levels-of-Extremists-using-Tweets
byRESHAN FARAZ
 
Data Science in the Real World: Making a Difference
bySrinath Perera
 
LOD2 Webinar Series FOX
byLOD2 Creating Knowledge out of Interlinked Data
 
The science behind predictive analytics a text mining perspective
byankurpandeyinfo
 
CityPulse - Wright State University
byCityPulse Project
 
A Quick Overview of Artificial Intelligence and Machine Learning
byHiroki Sayama
 
ppt
bybutest
 
Data: Past, Present, and Future (Lecture 1, Spring 2018)
bychris wiggins
 
ESWC SS 2012 - Friday Keynote Marko Grobelnik: Big Data Tutorial
byeswcsummerschool
 

More from Heiko Paulheim

PDF
Knowledge Graph Generation from Wikipedia in the Age of ChatGPT: Knowledge ...
byHeiko Paulheim
 
PDF
What_do_Knowledge_Graph_Embeddings_Learn.pdf
byHeiko Paulheim
 
PDF
New Adventures in RDF2vec
byHeiko Paulheim
 
PDF
Knowledge Matters! The Role of Knowledge Graphs in Modern AI Systems
byHeiko Paulheim
 
PDF
From Wikis to Knowledge Graphs
byHeiko Paulheim
 
PDF
Using Knowledge Graphs in Data Science - From Symbolic to Latent Representati...
byHeiko Paulheim
 
PPT
Beyond DBpedia and YAGO – The New Kids on the Knowledge Graph Block
byHeiko Paulheim
 
ODP
From Wikipedia to Thousands of Wikis – The DBkWik Knowledge Graph
byHeiko Paulheim
 
ODP
Make Embeddings Semantic Again!
byHeiko Paulheim
 
ODP
How much is a Triple?
byHeiko Paulheim
 
ODP
Machine Learning with and for Semantic Web Knowledge Graphs
byHeiko Paulheim
 
ODP
Weakly Supervised Learning for Fake News Detection on Twitter
byHeiko Paulheim
 
PDF
Towards Knowledge Graph Profiling
byHeiko Paulheim
 
ODP
Knowledge Graphs on the Web
byHeiko Paulheim
 
ODP
Data-driven Joint Debugging of the DBpedia Mappings and Ontology
byHeiko Paulheim
 
ODP
Fast Approximate A-box Consistency Checking using Machine Learning
byHeiko Paulheim
 
PPT
Serving DBpedia with DOLCE - More Than Just Adding a Cherry on Top
byHeiko Paulheim
 
ODP
Combining Ontology Matchers via Anomaly Detection
byHeiko Paulheim
 
PPT
Gathering Alternative Surface Forms for DBpedia Entities
byHeiko Paulheim
 
ODP
What the Adoption of schema.org Tells about Linked Open Data
byHeiko Paulheim
 
Knowledge Graph Generation from Wikipedia in the Age of ChatGPT: Knowledge ...
byHeiko Paulheim
 
What_do_Knowledge_Graph_Embeddings_Learn.pdf
byHeiko Paulheim
 
New Adventures in RDF2vec
byHeiko Paulheim
 
Knowledge Matters! The Role of Knowledge Graphs in Modern AI Systems
byHeiko Paulheim
 
From Wikis to Knowledge Graphs
byHeiko Paulheim
 
Using Knowledge Graphs in Data Science - From Symbolic to Latent Representati...
byHeiko Paulheim
 
Beyond DBpedia and YAGO – The New Kids on the Knowledge Graph Block
byHeiko Paulheim
 
From Wikipedia to Thousands of Wikis – The DBkWik Knowledge Graph
byHeiko Paulheim
 
Make Embeddings Semantic Again!
byHeiko Paulheim
 
How much is a Triple?
byHeiko Paulheim
 
Machine Learning with and for Semantic Web Knowledge Graphs
byHeiko Paulheim
 
Weakly Supervised Learning for Fake News Detection on Twitter
byHeiko Paulheim
 
Towards Knowledge Graph Profiling
byHeiko Paulheim
 
Knowledge Graphs on the Web
byHeiko Paulheim
 
Data-driven Joint Debugging of the DBpedia Mappings and Ontology
byHeiko Paulheim
 
Fast Approximate A-box Consistency Checking using Machine Learning
byHeiko Paulheim
 
Serving DBpedia with DOLCE - More Than Just Adding a Cherry on Top
byHeiko Paulheim
 
Combining Ontology Matchers via Anomaly Detection
byHeiko Paulheim
 
Gathering Alternative Surface Forms for DBpedia Entities
byHeiko Paulheim
 
What the Adoption of schema.org Tells about Linked Open Data
byHeiko Paulheim
 

Recently uploaded

PDF
Bettersize | BeSEC Series Product Brochure
byBettersize Instruments
 
PDF
Quick Learn Laravel for Beginner from Zero to Survive
byiDev Semarang
 
PDF
Parental Control App for Phones_ The Complete 2026 Guide for Safer, Smarter P...
byRyan Cooper
 
PDF
How to Design Premium Health (Public Health) PPT Using AI? Top Strategies
byPublic Health Concern Nepal
 
PDF
Afsar Ahmed - Digital Marketing Portfolio
byAfsar Ahmed
 
PDF
"Painless Major Upgrade: A Strategy for Updating Large Codebases", Andrii Yat...
byFwdays
 
PDF
Taxonomy Alignment for SDG Tagging - ASHA Case Study
byEnterprise Knowledge
 
PDF
When Drones Decide for Themselves_ Inside the Rise of Machine-Led Flight.pdf
byLyra Anderson
 
PDF
Transcript: EU regulations for the North American book supply chain - Tech Fo...
byBookNet Canada
 
PDF
Real-Time AI Masterclass: Vector Search at Scale
byScyllaDB
 
PDF
Transcript: What Thema can do: Leveraging metadata to support the discoverabi...
byBookNet Canada
 
PDF
JCB_3CX_SPECS (24 págs) companywrench.com .pdf
byAquilesOyarzn1
 
PDF
Fortinet Certified Fundamentals in Cybersecurity
byVICTOR MAESTRE RAMIREZ
 
PDF
Pneumatic Pressure Pump PPP01 for Calibration & Testing
bySERRAX TECHNOLOGIES LLP
 
PDF
Computer-Based Training (CBT) The Backbone of Modern Technical & Defence Trai...
bycomputerbasedtrainin1
 
PDF
Dev Dives: Build and deploy agentic automations - the unified way
byUiPathCommunity
 
PDF
Enterprise Data Services_ A Development Guide with Use Cases, Trends and Impl...
by2601harsh23
 
PDF
Adapt.com Seed Fundraising Deck | The AI Computer for Business
byashley459565
 
PDF
A Simple Guide to Real Estate Tokenization on XRP Ledger.pdf
byemmajoh2025
 
PPTX
INSY_7213_Theme Sessions 47 & 48 Advanced databases.pptx
bystormzy56
 
Bettersize | BeSEC Series Product Brochure
byBettersize Instruments
 
Quick Learn Laravel for Beginner from Zero to Survive
byiDev Semarang
 
Parental Control App for Phones_ The Complete 2026 Guide for Safer, Smarter P...
byRyan Cooper
 
How to Design Premium Health (Public Health) PPT Using AI? Top Strategies
byPublic Health Concern Nepal
 
Afsar Ahmed - Digital Marketing Portfolio
byAfsar Ahmed
 
"Painless Major Upgrade: A Strategy for Updating Large Codebases", Andrii Yat...
byFwdays
 
Taxonomy Alignment for SDG Tagging - ASHA Case Study
byEnterprise Knowledge
 
When Drones Decide for Themselves_ Inside the Rise of Machine-Led Flight.pdf
byLyra Anderson
 
Transcript: EU regulations for the North American book supply chain - Tech Fo...
byBookNet Canada
 
Real-Time AI Masterclass: Vector Search at Scale
byScyllaDB
 
Transcript: What Thema can do: Leveraging metadata to support the discoverabi...
byBookNet Canada
 
JCB_3CX_SPECS (24 págs) companywrench.com .pdf
byAquilesOyarzn1
 
Fortinet Certified Fundamentals in Cybersecurity
byVICTOR MAESTRE RAMIREZ
 
Pneumatic Pressure Pump PPP01 for Calibration & Testing
bySERRAX TECHNOLOGIES LLP
 
Computer-Based Training (CBT) The Backbone of Modern Technical & Defence Trai...
bycomputerbasedtrainin1
 
Dev Dives: Build and deploy agentic automations - the unified way
byUiPathCommunity
 
Enterprise Data Services_ A Development Guide with Use Cases, Trends and Impl...
by2601harsh23
 
Adapt.com Seed Fundraising Deck | The AI Computer for Business
byashley459565
 
A Simple Guide to Real Estate Tokenization on XRP Ledger.pdf
byemmajoh2025
 
INSY_7213_Theme Sessions 47 & 48 Advanced databases.pptx
bystormzy56
 

Exploiting Linked Open Data as Background Knowledge in Data Mining

  • 1.
    10/08/13 Heiko Paulheim1 Exploiting Linked Open Data as Background Knowledge in Data Mining Heiko Paulheim, University of Mannheim
  • 2.
    10/08/13 Heiko Paulheim2 Outline • Motivation • The original FeGeLOD framework • Experiments • Applications • The RapidMiner Linked Open Data Extension • Challenges and Future Work
  • 3.
    10/08/13 Heiko Paulheim3 Motivation: An Example Data Mining Task • Analyzing book sales ISBN City Sold 3-2347-3427-1 Darmstadt 124 3-43784-324-2 Mannheim 493 3-145-34587-0 Roßdorf 14 ... ISBN City Population ... Genre Publisher ... Sold 3-2347-3427-1 Darm- stadt 144402 ... Crime Bloody Books ... 124 3-43784-324-2 Mann- heim 291458 … Crime Guns Ltd. … 493 3-145-34587-0 Roß- dorf 12019 ... Travel Up&Away ... 14 ... → Crime novels sell better in larger cities
  • 4.
    10/08/13 Heiko Paulheim4 Motivation • Many data mining problems are solved better – when you have more background knowledge (leaving scalability aside) • Problems: – Tedious work – Selection bias: what to include?
  • 5.
    10/08/13 Heiko Paulheim5 Motivation http://lod-cloud.net/
  • 6.
    10/08/13 Heiko Paulheim6 Motivation • Idea: – reuse background knowledge from Linked Open Data – include it in the data mining process as needed • Two main variants: – develop mining/learning algorithms that run directly on Linked Data – create relational features from Linked Data
  • 7.
    10/08/13 Heiko Paulheim7 Motivation • Develop mining/learning algorithms – e.g., DL Learner – e.g., dedicated Kernel functions • Advantages: – can be quite efficient – no reduction to “flat” table structure – semantics can be respected directly
  • 8.
    10/08/13 Heiko Paulheim8 Motivation • Create relational features – e.g., LiDDM – e.g., AutoSPARQL – e.g., FeGeLOD / RapidMiner Linked Open Data Extension • Advantages: – Easy combination of knowledge from various sources • including relational features in the original data – Arbitrary mining algorithms/tools possible
  • 9.
    10/08/13 Heiko Paulheim9 FeGeLOD – Feature Generation from LOD IS B N 3 -2 3 4 7 -3 4 2 7 -1 C ity D a r m s ta d t # s o ld 1 2 4 N a m e d E n t it y R e c o g n it io n IS B N 3 -2 3 4 7 -3 4 2 7 - 1 C ity D a r m s ta d t # s o ld 1 2 4 C ity _ U R I h ttp : / / d b p e d ia .o r g / r e s o u r c e/ D a r m s ta d t F e a t u r e G e n e r a t io n IS B N 3 -2 3 4 7 -3 4 2 7 -1 C ity D a r m s ta d t # s o ld 1 2 4 C ity _ U R I h ttp : / / d b p e d ia .o r g / r e s o u r c e / D a r m s ta d t C ity _ U R I_ d b p e d ia -o w l: p o p u la tio n T o ta l 1 4 1 4 7 1 C ity _ U R I_ ... ... F e a t u r e S e le c t io n IS B N 3 -2 3 4 7 -3 4 2 7 - 1 C ity D a r m s ta d t # s o ld 1 2 4 C ity _ U R I h ttp : / / d b p e d ia .o r g / r e s o u r c e/ D a r m s ta d t C ity _ U R I_ d b p e d ia -o w l:p o p u la tio n T o ta l 1 4 1 4 7 1
  • 10.
    10/08/13 Heiko Paulheim10 FeGeLOD – Feature Generation from LOD • Original prototype, based on Weka: – Simple NER (guessing URIs) – Seven generators: • direct types • data properties • unqualified relations (boolean, numeric) • qualified relations (boolean, numeric) • individuals (dangerous!) - may be restricted to specific property – Simple feature selection: filtering features • that have only* different values (expect numerical) • that have only* identical values • that are mostly missing* *) 95% or 99%
  • 11.
    10/08/13 Heiko Paulheim11 Experiments • Testing with two* standard machine learning data sets – Zoo: classifying animals – AAUP: predicting income of university employees (regression task) • Question: how much improvement do additional features bring? *) standard ML datasets with speaking labels are scarce!
  • 12.
    10/08/13 Heiko Paulheim12 Experiments: Zoo Dataset
  • 13.
    10/08/13 Heiko Paulheim13 First Results: AAUP
  • 14.
    10/08/13 Heiko Paulheim14 Experiments: Early Insights • Additional features often improve the results • Zoo dataset: – Ripper: 89.11 to 96.04 – SMO: 93.07 to 97.03 – No improvement for Naive Bayes • AAUP dataset (compensation): – M5: 59.88 to 51.28 – SMO: 74.12 to 61.97 – No improvement for linear regression • ...but they may also cause problems – extreme example: 6.54 to 189.90 for linear regression – memory and timeouts due to large datasets
  • 15.
    10/08/13 Heiko Paulheim15 Experiments: Quality of Features • Information gain of features on Zoo dataset
  • 16.
    10/08/13 Heiko Paulheim16 Experiments: Quality of Features • Information gain of features on AAUP dataset (compensation)
  • 17.
    10/08/13 Heiko Paulheim17 Application: Classifying Events from Wikipedia • Event Extraction from Wikipedia • Joint work with Dennis Wegener and Daniel Hienert (GESIS) • Task: event classification (e.g., Politics, Sports, ...) http://www.vizgr.org/historical-events/timeline/
  • 18.
    10/08/13 Heiko Paulheim18 Application: Classifying Events from Wikipedia • Source Material: http://www.vizgr.org/historical-events/timeline/
  • 19.
    10/08/13 Heiko Paulheim19 Application: Classifying Events from Wikipedia • Positive Examples for class politics: – 2011, March 15 - German chancellor Angela Merkel shuts down the seven oldest German nuclear power plants. – 2010, June 3 – Christian Wulff is nominated for President of Germany by Angela Merkel. • Negative Examples for class politics: – 2010, July 7 – Spain defeats Germany 1-0 to win its semi-final and for its first time, along with Netherlands make the 2010 FIFA World Cup Final. – 2012, February 16 – Roman Lob is selected to represent Germany in the Eurovision Song Contest.
  • 20.
    10/08/13 Heiko Paulheim20 Application: Classifying Events from Wikipedia • Positive Examples for class politics: – 2011, March 15 - German chancellor Angela Merkel shuts down the seven oldest German nuclear power plants. – 2010, June 3 – Christian Wulff is nominated for President of Germany by Angela Merkel. • Negative Examples for class politics: – 2010, July 7 – Spain defeats Germany 1-0 to win its semi-final and for its first time, along with Netherlands make the 2010 FIFA World Cup Final. – 2012, February 16 – Roman Lob is selected to represent Germany in the Eurovision Song Contest. • Possible learned model: – "Angela Merkel" → Politics
  • 21.
    10/08/13 Heiko Paulheim21 Application: Classifying Events from Wikipedia • Possibly Learned Model: – "Angela Merkel" → Politics • How can we do better? • Background knowledge from Linked Open Data – 2011, March 15 - German chancellor Angela Merkel [class: Politician] shuts down the seven oldest German nuclear power plants. – 2012, May 13, Elections in North Rhine-Westphalia – Hannelore Kraft [class: Politician] is elected to continue as Minister-President, heading an SPD- Green coalition. • Model learned in that case: – "[class: Politician]" → Politics
  • 22.
    10/08/13 Heiko Paulheim22 Application: Classifying Events from Wikipedia • Model learned in that case: – "[class: Politician]" → Politics • Much more general – Can also classify events with politicians not contained in the training set • Less training examples required – A few events with politicians, athletes, singers, ... are enough
  • 23.
    10/08/13 Heiko Paulheim23 Application: Classifying Events from Wikipedia • Experiments on Wikipedia data – >10 categories – 1,000 labeled examples as training set – Classification accuracy: 80% • Plus: – We have trained a language-independent model! • often, models are like "elect*" → Politics – 22. Mai 2012: Peter Altmaier [class: Politician] wird als Nachfolger von Norbert Röttgen [class: Politician] zum Bundesumweltminister ernannt. – 6 januari 2012: Jonas Sjöstedt [class: Politician] väljs till ny partiledare för Vänsterpartiet efter Lars Ohly [class: Politician].
  • 24.
    10/08/13 Heiko Paulheim24 Application: Classifying Tweets • Joint work with Axel Schulz and Petar Ristoski (SAP Research) • Goal: using Twitter for emergency management fire at #mannheim #universityomg two cars on fire #A5 #accident fire at train station still burning my heart is on fire!!!come on baby light my fire boss should fire that stupid moron
  • 25.
    10/08/13 Heiko Paulheim25 Application: Classifying Tweets • Social media contains data on many incidents – But keyword search is not enough – Detecting small incidents is hard – Manual inspection is too expensive (and slow) • Machine learning could help – Train a model to classify incident/non incident tweets – Apply model for detecting incident related tweets • Training data: – Traffic accidents – ~2,000 tweets containing relevant keywords (“car”, “crash”, etc.), hand labeled (50% related to traffic incidents)
  • 26.
    10/08/13 Heiko Paulheim26 Application: Classifying Tweets • Learning to classify tweets: – Positive and negative examples – Features: • Stemming • POS tagging • Word n-grams • … • Accuracy ~90% • But – Accuracy drops to ~85% when applying the model to a different city
  • 27.
    10/08/13 Heiko Paulheim27 Application: Classifying Tweets • Example set: – “Again crash on I90” – “Accident on I90” • Model: – “I90” → indicates traffic accident • Applying the model: – “Two cars crashed on I51” → not related to traffic accident
  • 28.
    10/08/13 Heiko Paulheim28 Using LOD for Preventing Overfitting • Example set: – “Again crash on I90” – “Accident on I90” dbpedia:Interstate_90 dbpedia-owl:Road rdf:type dbpedia:Interstate_51 rdf:type • Model: – dbpedia-owl:Road → indicates traffic accident • Applying the model: – “Two cars crashed on I51” → indicates traffic accident • Using DBpedia Spotlight + FeGeLOD – Accuracy keeps up at 90% – Overfitting is avoided
  • 29.
    10/08/13 Heiko Paulheim29 Explaining Statistics • Statistics are very wide spread – Quality of living in cities – Corruption by country – Fertility rate by country – Suicide rate by country – Box office revenue of films – ...
  • 30.
    10/08/13 Heiko Paulheim30 Explaining Statistics • Questions we are often interested in – Why does city X have a high/low quality of living? – Why is the corruption higher in country A than in country B? – Will a new film create a high/low box office revenue? • i.e., we are looking for – explanations – forecasts (e.g., extrapolations)
  • 31.
    10/08/13 Heiko Paulheim31 Explaining Statistics http://xkcd.com/605/
  • 32.
    10/08/13 Heiko Paulheim32 Explaining Statistics • What statistics often look like
  • 33.
    10/08/13 Heiko Paulheim33 Explaining Statistics • There are powerful tools for finding correlations etc. – but many statistics cannot be interpreted directly – background knowledge is missing • Approach: – use Linked Open Data for enriching statistical data (e.g., FeGeLOD) – run analysis tools for finding explanations
  • 34.
    10/08/13 Heiko Paulheim34 Prototype Tool: Explain-a-LOD • Loads a statistics file (e.g., CSV) • Adds background knowledge • Runs basic analysis (correlation, rule learning) • Presents explanations
  • 35.
    10/08/13 Heiko Paulheim35 Statistical Data: Examples • Data Set: Mercer Quality of Living – Quality of living in 216 cities word wide – norm: NYC=100 (value range 23-109) – As of 1999 – http://across.co.nz/qualityofliving.htm • LOD data sets used in the examples: – DBpedia – CIA World Factbook for statistics by country
  • 36.
    10/08/13 Heiko Paulheim36 Statistical Data: Examples • Examples for low quality cities – big hot cities (junHighC >= 27 and areaTotalKm >= 334) – cold cities where no music has ever been recorded (recordedIn_in = false and janHighC <= 16) – latitude <= 24 and longitude <= 47 • a very accurate rule • but what's the interpretation? Next Record Studio 2547 miles Next Record Studio 2547 miles
  • 37.
    10/08/13 Heiko Paulheim37 Statistical Data: Examples
  • 38.
    10/08/13 Heiko Paulheim38 Statistical Data: Examples • Data Set: Transparency International – 177 Countries and a corruption perception indicator (between 1 and 10) – As of 2010 – http://www.transparency.org/cpi2010/results
  • 39.
    10/08/13 Heiko Paulheim39 Statistical Data: Examples • Example rules for countries with low corruption – HDI > 78% • Human Development Index, calculated from live expectancy, education level, economic performance – OECD member states – Foundation place of more than nine organizations – More than ten mountains – More than ten companies with their headquarter in that state, but less than two cargo airlines
  • 40.
    10/08/13 Heiko Paulheim40 Statistical Data: Examples • Data Set: Burnout rates – 16 German DAX companies – Absolute and relative numbers – As of 2011 – http://de.statista.com/statistik/daten/studie/226959/umfrage/burn-out- erkrankungen-unter-mitarbeitern-ausgewaehlter-dax-unternehmen/
  • 41.
    10/08/13 Heiko Paulheim41 Datavalues Type Unqualifiedrelation(boolean) Unqualifiedrelation(numeric) Qualifiedrelation(boolean) Qualifiedrelation(numeric) Joint 1 1.5 2 2.5 3 3.5 4 4.5 5 Correlation Rule Learning Evaluation of Feature Quality • Quality of living dataset
  • 42.
    10/08/13 Heiko Paulheim42 Datavalues Type Unqualifiedrelation(boolean) Unqualifiedrelation(numeric) Qualifiedrelation(boolean) Qualifiedrelation(numeric) Joint 1 1.5 2 2.5 3 3.5 4 4.5 5 Correlation Rule Learning Evaluation of Feature Quality • Corruption dataset
  • 43.
    10/08/13 Heiko Paulheim43 Statistical Data: Examples • Findings for burnout rates – Positive correlation between turnover and burnout rates – Car manufacturers are less prone to burnout – German companies are less prone to burnout than international ones • Exception: Frankfurt
  • 44.
    10/08/13 Heiko Paulheim44 Statistical Data: Examples • Data Set: Antidepressives consumption – In European countries – Source: OECD – http://www.oecd-ilibrary.org/social-issues-migration-health/health-at-a-glance- 2011/pharmaceutical-consumption_health_glance-2011-39-en
  • 45.
    10/08/13 Heiko Paulheim45 Statistical Data: Examples • Findings for antidepressives consumption – Larger countries have higher consumption – Low HDI → high consumption – By geography: • Nordic countries, countries at the Atlantic: high • Mediterranean: medium • Alpine countries: low – High average age → high consumption – High birth rates → high consumption
  • 46.
    10/08/13 Heiko Paulheim46 Statistical Data: Examples • Data Set: Suicide rates – By country – OECD states – As of 2005 – http://www.washingtonpost.com/wp-srv/world/suiciderate.html
  • 47.
    10/08/13 Heiko Paulheim47 Statistical Data: Examples • Findings for suicide rates – Democraties have lower suicide rates than other forms of government – High HDI → low suicide rate – High population density → high suicide rate – By geography: • At the sea → low • In the mountains → high – High Gini index → low suicide rate • High Gini index ↔ unequal distribution of wealth – High usage of nuclear power → high suicide rates
  • 48.
    10/08/13 Heiko Paulheim48 Statistical Data: Examples • Data set: sexual activity – Percentage of people having sex weekly – By country – Survey by Durex 2005-2009 – http://chartsbin.com/view/uya
  • 49.
    10/08/13 Heiko Paulheim49 Statistical Data: Examples • Findings on sexual activity – By geography: • High in Europe, low in Asia • Low in Island states – By language: • English speaking: low • French speaking: high – Low average age → high activity – High GDP per capita → low activity – High unemployment rate → high activity – High number of ISP providers → low activity
  • 50.
    10/08/13 Heiko Paulheim50 Try it... but be careful! • Download from http://www.ke.tu-darmstadt.de/resources/explain-a-lod • including a demo video, papers, etc. http://xkcd.com/552/
  • 51.
    10/08/13 Heiko Paulheim51 RapidMiner Linked Open Data Extension • August 16th , 2013: FeGeLOD celebrates its 2nd birthday • Problems – still no nice UI – special configurations are tricky – difficult to enhance • Decision – Reimplementation on RapidMiner platform – September 13th , 2013: Release of RapidMiner Linked Open Data Extension – Available from RapidMiner marketplace • http://dws.informatik.uni-mannheim.de/en/research/rapidminer-lod-extension/
  • 52.
    10/08/13 Heiko Paulheim52 RapidMiner Linked Open Data Extension • Simple wiring of operators – linkers – generators • Combination with powerful RapidMiner operators
  • 53.
    10/08/13 Heiko Paulheim53 RapidMiner Linked Open Data Extension • Easy SPARQL endpoint definitions • Support of custom SPARQL statements
  • 54.
    10/08/13 Heiko Paulheim54 Challenges and Future Work • SPARQL variants – Some endpoints support special/non-standard SPARQL constructs – COUNT(...) – transitive closure – exploit where applicable • Implementations without SPARQL – Freebase – OpenCyc
  • 55.
    10/08/13 Heiko Paulheim55 Challenges and Future Work • Linking is still challenging – URI patterns are not flexible – Search by label is time consuming – Services like DBpedia Lookup are scarce • Limitations of completely unsupervised linking – e.g., Hurricanes – how to use headlines/attribute names?
  • 56.
    10/08/13 Heiko Paulheim56 Challenges and Future Work • Linking as optimization problem – find candidates for all entities, e.g., by DBpedia lookup – find a selection of candidates that are most similar to each other • e.g., all of them are U.S. cities – some experiments with types and categories • problem: not complete – some problems cannot be addressed (e.g.: Hurricanes) • Alternatives: – semi supervised linking – user provides some example links – active learning
  • 57.
    10/08/13 Heiko Paulheim57 Challenges and Future Work • Exploiting semantics for feature selection • Given two features: – f1: type(RoadsInAlaska) – f2: type(Road) • and the schema definition Road rdfs:subclassOf RoadsInAlaska • Exploit that information for feature selection – e.g., gain(f1) ≈ gain(f2), f1<f2 → remove f1
  • 58.
    10/08/13 Heiko Paulheim58 Challenges and Future Work • Incompleteness of LOD – e.g., type information in DBpedia – may lead to findings such as • if a city is of type Place, the quality of living is high – possible remedy: autocomplete on the dataset (e.g., Paulheim/Bizer 2013) • Biases in LOD – e.g., DBpedia has a bias towards western culture – may lead to findings such as • if many records have been made in a city, the quality of living is high
  • 59.
    10/08/13 Heiko Paulheim59 Challenges and Future Work • Features not used for scalability reasons: – features for single entities • e.g., “Roman Polanski directorOf X” – features more than one hop away • e.g., “Cities with a university which has a computer science department” – some are covered by YAGO types, e.g., “AustralianBandsFoundedIn1990” • but subject to YAGO's selection bias • Approaches are required to use such features – which respect scalability – “generate first, filter later” is not the best solution • e.g., “Cities with at least one of ArtSchoolsInParis” – on-the-fly filtering may be more suitable • e.g., sampling
  • 60.
    10/08/13 Heiko Paulheim60 Challenges and Future Work • Automatically exploit data sources with non-simple structures EU18931 a Funding . EU18931 has-grant-value [ has-amount 1300000 . has-unit-of-measure EUR . ] • Support geo/temporal features – e.g., Data Cubes – e.g., Linked Geo Data • Construct complex features (in a scalable way!) – e.g., cinemas per inhabitant real example from CORDIS dataset
  • 61.
    10/08/13 Heiko Paulheim61 Wrap-up • Linked Data is useful as background knowledge – especially on problems which have little knowledge in themselves • Unsupervised methods – avoid biases and work without knowledge about LOD – but: scalability and generality problems • RapidMiner LOD extension – a constantly growing toolkit
  • 62.
    10/08/13 Heiko Paulheim62 Credits & Thanks • Past contributors of FeGeLOD: – Johannes Fürnkranz – Raad Bahmani – Alexander Gabriel – Simon Holthausen • Current team of RapidMiner Linked Open Data Extension: – Chris Bizer – Petar Ristoski – Evgeny Mitichkin
  • 63.
    10/08/13 Heiko Paulheim63 Exploiting Linked Open Data as Background Knowledge in Data Mining Heiko Paulheim, University of Mannheim