Using the Semantic Web Stack to Make Big Data Smarter

Matheus Mota
Doutorando @UNICAMP

Transcript

1. Using the Semantic Web Stack to Make Big Data Smarter Matheus Mota PhD Candidate @ LIS.IC.UNICAMP @matheusmota /msmota matheusmota.com
2. “ Semantic web is an extension of the WWW that enables both sharing and integration of content beyond the boundaries of applications and websites" http://www.dataversity.net/big-data-semantic-web-technology-data-visualization Semantic Web Adapted from DataVersity 2
3. “ It is all about structured extra stuff that you should put under web's hood to feed our robots". http://www.dataversity.net/big-data-semantic-web-technology-data-visualization Semantic Web 3
4. http://www.dataversity.net/big-data-semantic-web-technology-data-visualization Semantic Web 4
5. http://www.dataversity.net/big-data-semantic-web-technology-data-visualization Semantic Web 5
6. http://www.dataversity.net/big-data-semantic-web-technology-data-visualization Semantic Web 6
7. http://www.dataversity.net/big-data-semantic-web-technology-data-visualization Semantic Web 7
8. “ big data is going to give to semantic web the massive amounts of metadata it needs to really get traction." http://radar.oreilly.com/2011/06/big-data-and-the-semantic-web.html Big Semantic Web of Data Edd Dumbill 8
9. WWW yesterday 9
10. WWW as a platform 10
11. “ big data is going to give to semantic web the massive amounts of metadata it needs to really get traction." http://radar.oreilly.com/2011/06/big-data-and-the-semantic-web.html Big Semantic Web of Data Edd Dumbill 11
12. http://radar.oreilly.com/2011/06/big-data-and-the-semantic-web.html Semantic Web: (new?) players Source: [3] 12
13. The so-called Semantic Web Cake You won't need it all to bootstrap 13
14. The so-called Semantic Web Stack 14
15. The so-called Semantic Web Stack Future Present Past 15
16. Semantic Web - The Goal 16
17. Getting Started Sounds cool. How can I play? 17
18. “ -Flexible Modeling for interconnected data -Agile Evolution of the Data Model -Scalable Evaluation of Join-Intensive Queries/Paths It all starts with graphs". Is it good for your problem/scenario/data? 18
19. 19 Types of Graph-related tasks - Link Prediction
20. 20 Types of Graph-related tasks - Link Prediction ? ? ? ?
21. 21 Topology aware Link Prediction 1 4 4 3
22. 22 1 4 4 3 <new_user> <famousEnt> Topology aware Link Prediction
23. 23 Topology Agnostic - Properties/Semantics
24. 24 sameAs rdf:type Topology Agnostic - Properties/Semantics
25. Semantic Web vs. Linked Data 25
26. 26 LinkedData Semantic Web vs. Linked Data
27. DBPedia 27
28. 28 DBPedia
29. 29 DBPedia
30. Semantic Web - The Goal 30
31. Semantic Web - The Goal 31
32. Publishing/ Querying How to? P/Q 32
33. Publishing: HTML Embedded -Microformats -Microdata -RDFa -N3-in-HTML 33
34. It is already out there 34
35. It is already out there 35
36. It is already out there http://radar.oreilly.com/2011/06/big-data-and-the-semantic-web.html 36
37. P/Q: Data (SPARQL) Endpoints -OpenLink Virtuoso -Allegro Graph -Apache Jena + SQL BD -GraphDB -Adhoc Interface + another graph DB (neo4j, etc) 37
38. P/Q: Data (SPARQL) Endpoints 38
39. P/Q: Main Challenges 39 - ETL - Performance - Motivation =(
40. Going Further What is next? 40
41. Ontology/ Inference Modeling and Inferences O/I 41
42. Modeling RDF Triple Structure. RDFs Adds taxonomic relations, object relations and datatype properties. OWL Richer vocabulary; cardinality, logical operators; semantic relationships between objects. 42
43. RDF/ RDFs 43 ns:motorV ns:SUV rdfs:subClassOf rdf:type rdf:type a rdfs:Class
44. RDF/ RDFs 44 ns2:X_33 ns2:seller 1 ns:sold ns:motorV ns:SUV rdfs:subClassOf rdf:type rdf:type a rdfs:Class rdfs:range
45. RDF/ RDFs / OWL 45 ns2:x_33 ns2:seller 1 ns:measured ns:4wheeler ns:SUV rdfs:subClassOf rdf:type rdf:type a rdfs:Class rdfs:range ns:motorV rdfs:subClassOf rdf:type a
46. RDF/ RDFs / OWL 46 ns2:x_33 ns2:seller 1 ns:measured ns:4wheeler ns:SUV rdfs:subClassOf rdf:type rdf:type a rdfs:Class rdfs:range ns:motorV rdfs:subClassOf ns:2wheeler rdf:type a owl:disjoint
47. RDF/ RDFs / OWL 47
49. ns:4wheeler ns:SUV rdfs:subClassOf rdf:type rdf:type a rdfs:Class ns:motorV rdfs:subClassOf ns:2wheeler rdf:type a owl:disjoint Inferences
50. ns:4wheeler ns:SUV rdfs:subClassOf rdf:type rdf:type a rdfs:Class ns:motorV rdfs:subClassOf ns:2wheeler rdf:type a owl:disjoint rdfs:subClassOf Inferences
51. ns:van ns:4wheeler rdfs:subClassOf ns:kombi rdf:type rdf:type rdf:type a rdfs:Class Inferences
52. ns:van ns:4wheeler rdfs:subClassOf ns:kombi rdf:type rdf:type rdf:type a rdfs:Class rdfs:type Inferences
53. Research @ LIS Powered by LinkedData 53
56. • • 56
57. • • 57 Clustering Similarity Det. Copy detection Ranking
59. 59 Endpoint: http://sparql.lis.ic.unicamp.br Graph: http://lis.ic.unicamp.br/fnjv
62. Scenario 62
63. 63 Mental-Map - Pattern on Creation SS Classification event, Proc., Object
65. http://purl.org/biospread 65
66. Data Endpoint: http://sparql.lis.ic.unicamp.br Project: http://purl.org/biospread/ 66
67. Thanks! Any questions? 67 @matheusmota /msmota matheusmota.com mota@ic.unicamp.br @
68. Acknowledgments Credits Special Thanks to ◎ Professor André Santanchè (including slides) ◎ Laboratory of Information Systems (http://lis.ic.unicamp.br) ◎ Institute Of Computing - UNICAMP Thanks to all the people who made and released these awesome resources for free: ◎ Presentation template by SlidesCarnival ◎ Photographs by Unsplash & Death to the Stock Photo (license) 68