How to be one step ahead of leveraging knowledge technologies for your apps!
When: Dec 8, 2017
Where: Fl. 6, Multimedia Tower, Central Jakarta
Thanks to Ragil for the invitation!
2. Fariz Darari
• 1988: Born in Malang
• 2010: BSc in Computer Science at Universitas Indonesia
• 2013: MSc in Computational Logic at University of Bolzano,
Italy and TU Dresden, Germany
Best Thesis Award and Enno-Heidebroek Award
• 2017: PhD in Computational Logic at University of Bolzano,
Italy and TU Dresden, Germany
• 2017: Lecturer at Faculty of CS, Universitas Indonesia
6. • Knowledge Technologies: Motivations
• Semantic Web
• Knowledge Bases These Days (= Zaman Now)
• Wikidata
• DBpedia
• Applications
• Discussion: Challenges & Opportunities
Menu
7. What if the knowledge in your brains,*
can be queried by computers?
*notice the plural form
8. What if the knowledge in your brains,*
can be queried by computers?
*notice the plural form
9. What if the knowledge in your brains,*
can be queried by computers?
*notice the plural form
10. What if the knowledge in your brains,
can be queried by computers?
Can you imagine what kind of advancements
can be made to humanity?
11. What if the knowledge in your brains,
can be queried by computers?
Can you imagine what kind of advancements
can be made to humanity?
Stay tuned, will present you an answer to this question some slides later!
12. ... if properly designed,
the Semantic Web can assist
the evolution of human knowledge
as a whole.
– Tim Berners-Lee
Inventor of the (Semantic) Web
15. What is the Semantic Web?
The set of technologies to put knowledge on the Web,
that is based on the following four principles:
1. Use URIs (Universal Resource Identifiers)* for identifying things
2. Use HTTP** URIs so people can look up those names
3. When someone looks up a URI, provide useful knowledge
using the standards: RDF and SPARQL.
4. Include links to other URIs, so they can discover more things
https://www.w3.org/DesignIssues/LinkedData.html
* URI = just like URL (web address), but you use it to identify things just like barcode for supermarket stuff!
** HTTP = the mechanism you use every time you access the Web!
18. RDF in one slide
the data guy
• Data model, based on S-P-O triple structure (Subject, Predicate, Object)
• Used for describing things, yes, every, single, thing
And anyway, RDF = Resource Description Framework
• Key features:
• RDF data can be exported in JSON and XML
• RDF links things, not just documents
• RDF links are typed
TelkomUniversity somelink Bandung
TelkomUniversity locatedIn Bandung
19. OWL in one slide
• Schema (=Ontology) language, describing vocabularies
• Yes, it is on the meta-level!
• Short for: Web Ontology Language (WOL? No, it is OWL!)
• Key features:
• Reasoning: you can check if your knowledge is consistent/not!
• Reasoning again: you can conclude new things
based on existing facts.
• Very simple example:
owl SubClassOf bird + bird SubClassOf animal + owl EquivalentClass strigifomes
Now, if Bobi is a Strigifomes, do you think Bobi is an animal?
the schema guy
20. OWL in one slide
• Schema (=Ontology) language, describing vocabularies
• Yes, it is on the meta-level!
• Short for: Web Ontology Language (WOL? No, it is OWL!)
• Key features:
• Reasoning: you can check if your knowledge is consistent/not!
• Reasoning again: you can conclude new things
based on existing facts.
• Very simple example:
owl SubClassOf bird + bird SubClassOf animal + owl EquivalentClass strigifomes
Now, if Bobi is a Strigifomes, do you think Bobi is an animal? OWL will say:
the schema guy
21. OWL in one slide
• Schema (=Ontology) language, describing vocabularies
• Yes, it is on the meta-level!
• Short for: Web Ontology Language (WOL? No, it is OWL!)
• Key features:
• Reasoning: you can check if your knowledge is consistent/not!
• Reasoning again: you can conclude new things
based on existing facts.
• Very simple example:
owl SubClassOf bird + bird SubClassOf animal + owl EquivalentClass strigifomes
Now, if Bobi is a Strigifomes, do you think Bobi is an animal? OWL will say: "YES!"
the schema guy
22. SPARQL in one slide
the query guy
• Query language: If RDF captures knowledge,
SPARQL asks questions about knowledge!
• Short for: SPARQL Protocol and RDF Query Language
• Key features: Asking for knowledge, is a KEY feature!
• Very simple example:
TelkomUniversity locatedIn Bandung
Bandung headOfGov RidwanKamil
TelkomUniversity instanceOf University
It is SPARQLing!
SELECT ?university WHERE {
?university instanceOf University .
?university locatedIn ?city .
?city headOfGov RidwanKamil }
Guess what this query is asking for?
HINT: Question mark (?) represents variables to match
with RDF data!
25. Knowledge Bases (KBs) These Days (Zaman Now)
KB NOW
Subject
Predicate
Predicate
Predicate
Object
Object
Object
Reminds you of something?
26. Knowledge Bases (KBs) These Days (Zaman Now)
KB NOW
Subject
Predicate
Predicate
Predicate
Object
Object
Object
Reminds you of something?
the data guy
27. Knowledge Bases (KBs) These Days (Zaman Now)
KB NOW
Subject
Predicate
Predicate
Predicate
Object
Object
Object
Reminds you of something?
the data guy
btw, every subject in Wikidata has its own identifier, the URI is made by: Wikidata domain + identifier
28. Knowledge Bases (KBs) These Days (Zaman Now)
KB NOW
Subject
Predicate
Predicate
Predicate
Object
Object
Object
Reminds you of something?
the data guy
btw, every subject in Wikidata has its own identifier, the URI is made by: Wikidata domain + identifier
= P31
= P571
= Q4830453
= Q10389
29. Knowledge Bases (KBs) These Days (Zaman Now)
the query guy
http://tinyurl.com/yc6jsmhv
30. Knowledge Bases (KBs) These Days (Zaman Now)
the query guy
http://tinyurl.com/y84kyl4d
31. Knowledge Bases (KBs) These Days (Zaman Now)
the schema guy
OwlInWinnieThePooh instanceOf fictionalOwl
fictionalOwl subClassOf fictionalBird
fictionalBird subClassOf fictionalAnimal
32. Knowledge Bases (KBs) These Days (Zaman Now)
Wikidata key features:
• It is like Wikipedia but for data!
• It is under Wikimedia foundation
• It is crowdsourced, anyone can add data
• It is free
• It's got 326 million facts about 40 million
subjects! (Wikipedia only has 5 million subjects!)
• It loves the Semantic Web
34. Knowledge Bases (KBs) These Days (Zaman Now)
DBpedia key features:
• It extracts data from Wikipedia infoboxes
(summary box on top right corner).
• It is free
• It's got 13 BILLION facts about 7 million
subjects!
• It loves the Semantic Web
35. Knowledge Bases (KBs) These Days (Zaman Now)
DBpedia key features:
• It extracts data from Wikipedia infoboxes
(summary box on top right corner).
• It is free
• It's got 13 BILLION facts about 7 million
subjects!
• It loves the Semantic Web
• DBpedia Indonesia is available, hosted by
Faculty of Computer Science, Univ. Indonesia
42. Question: When was Soekarno born?
http://id.dbpedia.org/page/Soekarno
Application: DBpedia-powered Answer Engine
43. Application: DBpedia-powered Answer Engine
Question: When was Soekarno born?
Borrow Techniques from
Natural Language Processing
SELECT ?birthDate
WHERE {
<http://id.dbpedia.org/resource/Soekarno> <http://dbpedia.org/ontology/birthDate> ?birthDate
}
SPARQL Query over DBpedia
http://id.dbpedia.org/sparql
45. Application: Timeline Infographics
Task: Create timeline of Indonesian national heroes based on their birthdates!
Without Wikidata:
- Read by eyes websites about national heroes (there are all 173 heroes!)
- Gather information manually
- Visualize information manually
Total time spent: 24+ hours!
46. Application: Wikidata-powered Timeline Infographics
Task: Create timeline of Indonesian national heroes based on their birthdates!
With Wikidata (and Histropedia):
- Formulate and evaluate the query
- VOILA: Beautiful timeline infographics created!
Total time spent: 10 minutes!
55. Application: Wikidata-powered Virtual Doctor
dr Wikidata: Tell me your symptoms
Patient: I feel like fatigue, headache,
joint pain, and vomitting
dr Wikidata: From what I know,
you most likely get dengue fever!
65. Completeness: Is the data complete
enough? Is it of sufficient breadth and
depth?
Accuracy: How accurate is the data? Is
it reliable and verifiable?
Timeliness: Is the data up-to-date? Is
the latest data included?
Data Quality