Semantic Web Technologies: A Paradigm for Medical InformaticsChimezie OgbujiThe document discusses the application of semantic web technologies in medical informatics, highlighting challenges such as semantic interoperability and the integration of electronic health records. It emphasizes the importance of structured data, controlled vocabularies, and standard querying interfaces for achieving effective data exchange across clinical systems. Key technologies mentioned include RDF, OWL, and SPARQL, with examples from institutions like Cleveland Clinic and Mayo Clinic illustrating practical implementation.
Zarządzanie danymi badawczymiPlatforma Otwartej NaukiPrezentacja z warsztatów "Zarządzanie danymi badawczymi", które zostały poprowadzone przez Martę Hoffman-Sommer 9 grudnia 2015 r. i 11 grudnia 2015 r.
Marcin Roszkowski : Modelowanie uniwersum bibliograficznego, czyli do czego p...Zakład Systemów Informacyjnych, Instytut Informacji Naukowej i Studiów Bibliologicznych (UW)II Konferencja Naukowa : Nauka o informacji (informacja naukowa) w okresie zmian, Warszawa, 15-16.04.2013 r. Instytut Informacji Naukowej i Studiów Bibliologicznych, Uniwersytet Warszawski
The 2nd Scientific Conference : Information Science in an Age of Change, April 15-16, 2013. Institute of Information and Book Studies, University of Warsaw
Gene Wiki and Wikimedia Foundation SPARQL workshopBenjamin GoodThis document summarizes a presentation about curating biomedical knowledge on Wikidata and Wikipedia through the Gene Wiki project. The Gene Wiki project develops tools and resources to automatically generate gene pages on Wikipedia using structured data from Wikidata. This centralized biomedical knowledge on open platforms and allows the data to be queried through SPARQL, powering new applications for biomedical research.
Multilingual challenges for accessing digitized culture online - Riga Summit 15Antoine IsaacThe document discusses the multilingual challenges in accessing digitized cultural heritage online, highlighting the issues of data heterogeneity and the complexity of multilingual enrichment. It emphasizes the diversity of applications and the difficulties in applying language technology to effectively manage and utilize cultural data across Europe. The need for collaboration within the cultural heritage and language technology sectors is stressed, alongside the importance of open data and community sharing.
Modelling Knowledge Organization Systems and StructuresMarcia ZengThe document discusses the modeling of knowledge organization systems and structures within the context of conceptual and data models, particularly focusing on the FRAD conceptual model. It explains the relationships between 'thema' and 'nomen,' the attributes of these entities, and their implementation in data models like SKOS. The presentation illustrates how the FRAD model can fulfill the requirements of both traditional library systems and emerging linked data environments.
KOSO Knowledge Organization Systems OntologyKatrin WellerKOSO is a metadata ontology that aims to support knowledge exchange and reuse of existing knowledge organization systems (KOS) by providing descriptive metadata about different types of KOS, how they are classified and defined, and how they can interact. The ontology defines key concepts like KnowledgeOrganizationSystem and its subclasses like Ontology, Thesaurus, and Classification. It specifies properties to describe a KOS like its domain, language, and relations. It also models interactions between KOS through properties like has_version and is_interlinked_with. The goal is to enable discovery and understanding of existing KOS for improved reuse.
Multilingual Knowledge Organization Systems Management: Best PracticesMauro DragoniThe document discusses the modeling and management of multilingual Knowledge Organization Systems (KOS) in agriculture, highlighting their role in improving information access and reducing ambiguity. It outlines challenges such as multilinguality, collaborative work among experts, and the evolution of existing KOS, while introducing MOKI, a collaborative architecture using wikis and supporting multilingual ontology management. The Organic.Lingua project is mentioned, detailing its features, including connections to machine translation services and significant updates to the ontology during the project.
Wikipedia as source of collaboratively created Knowledge Organization SystemsJakob .The document discusses Wikipedia as a source of collaboratively created knowledge organization systems. It describes the structure of Wikipedia articles, categories, infoboxes, and how this structured data can be extracted and represented in semantic formats like RDF to create knowledge bases like DBpedia that link open data on the web. It also discusses some open issues around data quality, concepts and mapping when extracting and querying structured knowledge from Wikipedia.
Using OWL for the RESO Data DictionaryChimezie OgbujiThe document discusses the use of OWL (Ontology Web Language) for enhancing the RESO Data Dictionary to improve semantic interoperability and prevent misunderstandings in data communication. It outlines the structure of ontologies, including classes, properties, and relationships, and the benefits of using these for data representation in real estate contexts. Additionally, challenges and tools for implementing OWL in collaboration with domain experts are addressed, emphasizing the importance of well-designed ontologies for effective standardization.
DigiBird: on the fly collection integration using crowdsourcing VU University AmsterdamChris Dijkshoorn discusses challenges in crowdsourcing for data collection, including isolation, time demands, and integration into existing systems. The Digibird project proposes solutions like creating a hub for on-the-fly integration and using shared vocabularies to improve data access and collaboration. It emphasizes the need for continuous effort in promoting initiatives and refining contributions while acknowledging the complexities of developing a mature infrastructure.
Crowdsourcing Linked Data Quality AssessmentMaribel Acosta DeibeThe document presents a methodology for crowdsourcing the assessment of Linked Data quality. The methodology involves a two stage process - a find stage using Linked Data experts to identify potential quality issues, and a verify stage using microtasks on Amazon Mechanical Turk to validate the issues. The study assesses three types of quality problems in DBpedia through this methodology and analyzes the results in terms of precision. The findings indicate that crowdsourcing is effective for detecting certain quality issues, and that the expertise of Linked Data experts is best for domain-specific tasks while microtask workers perform well on data comparison tasks. The conclusions discuss integrating crowdsourcing into Linked Data curation processes and conducting further experiments.
Brief on Linked Data for U.S. EPA's Chief Data Scientist3 Round StonesThe document outlines the challenges and potential of data management at the EPA, highlighting the vast variety of data formats and silos that currently exist. It emphasizes the use of linked data and NoSQL platforms to improve data findability, accessibility, and interoperability, thus enhancing efficiency and transparency. Additionally, it discusses the implementation of a linked data management system to facilitate public access to environmental data and improve interactions across various datasets.
Role of Ontologies in Semantic Digital LibrariesSebastian Ryszard Kruk1) Ontologies play a key role in semantic digital libraries by supporting bibliographic descriptions, extensible resource structures, and community-aware features.
2) Semantic digital libraries integrate information from various metadata sources and provide interoperability between systems using semantics.
3) Key ontologies for digital libraries include bibliographic ontologies, structure description ontologies, and community-aware ontologies that model folksonomies and social semantic collaborative filtering.
Semantic Data Management in Graph Databases: ESWC 2014 TutorialMaribel Acosta DeibeThe document is a tutorial on semantic data management in graph databases, focusing on their structure and operations. It discusses graph data models, basic operations, various implementations, and the advantages of graph databases. Additionally, it covers existing graph database engines and their specific features for managing persistent and temporary graphs.
Linked Open Data in RomaniaVlad PoseaThis document discusses transforming open government data from Romania into linked open data. It begins with background on linked data and open data initiatives. Then it describes efforts to model, transform, link, and publish Romanian open data as linked open data. This includes identifying common vocabularies and properties, creating URIs, linking to external datasets like DBPedia, and publishing the linked data for use in applications via a SPARQL endpoint. Overall the goal is to make this data more accessible and interoperable through semantic web standards.
VIII Encuentros de Centros de Documentación de Arte Contemporáneo en Artium -...Artium VitoriaThe document outlines the development and vision for using linked data in cultural heritage on the semantic web from 2002 to 2016, emphasizing the creation of a collaborative framework for cultural institutions. It addresses the challenges of content complexity and production while presenting the 'sampo' model as a solution for semantic portals that facilitate access to rich cultural heritage resources. Three applications, including 'culturesampo' and 'warsampo', exemplify the effective use of linked data to enhance user engagement and research opportunities.
Wikidata and the Semantic Web of FoodBenjamin GoodThe document discusses the integration of Wikidata in building a semantic web of food knowledge, outlining its open and editable nature and its functionality as a knowledge base. It explains how Wikidata's structured items and statements create interconnected information, facilitating queries related to biomedical content, including genes, proteins, and diseases. The document also highlights various applications of this semantic web, supported by grants from the National Institute of Health.
Knowledge Organization System (KOS) for biodiversity information resources, G...Dag EndresenThe NCBO webinar focuses on the Global Biodiversity Information Facility (GBIF) and its work program aimed at enhancing the organization of biodiversity information. It emphasizes enabling free access to biodiversity data, and discusses the development of community standards, tools, and protocols for data management and publication. The presentation also outlines a governance structure for maintaining knowledge organization systems in biodiversity, along with recommendations for future developments in the field.
Knowledge Organization SystemsR A AkerkarThe document discusses knowledge organization systems (KOS) and how the Simple Knowledge Organization System (SKOS) bridges KOS and the Semantic Web. It provides examples of KOS like taxonomies and thesauruses and explains how they are used differently than ontologies. SKOS is defined as an RDF vocabulary for representing KOS online in a machine-readable way and became a W3C standard in 2009.
CHIST-ERA 2019 - presentation of CAMIL (Poznan University of Technology)Agnieszka ŁawrynowiczThe Center for Artificial Intelligence and Machine Learning (CAMIL) is a research organization in Poland that focuses on developing explainable machine learning systems and methodologies. Some of its areas of expertise include interpretable machine learning, decision support, active learning using explanations, and applying ontologies and knowledge graphs. It is interested in collaborating with others on projects involving using knowledge graphs and semantics for explainable AI.
Ontologie w historyczno-geograficznych systemach informacyjnychAgnieszka ŁawrynowiczPrezentacja z konferencji DARIAH-PL organizowanej przez PCSS. Wyniki analizy stanu badań w ramach projektu NPRH: Ontologiczne podstawy budowy historycznych systemów informacji geograficznej 2bH 15 0216 83
Modelling Knowledge Organization Systems and StructuresMarcia ZengThe document discusses the modeling of knowledge organization systems and structures within the context of conceptual and data models, particularly focusing on the FRAD conceptual model. It explains the relationships between 'thema' and 'nomen,' the attributes of these entities, and their implementation in data models like SKOS. The presentation illustrates how the FRAD model can fulfill the requirements of both traditional library systems and emerging linked data environments.
KOSO Knowledge Organization Systems OntologyKatrin WellerKOSO is a metadata ontology that aims to support knowledge exchange and reuse of existing knowledge organization systems (KOS) by providing descriptive metadata about different types of KOS, how they are classified and defined, and how they can interact. The ontology defines key concepts like KnowledgeOrganizationSystem and its subclasses like Ontology, Thesaurus, and Classification. It specifies properties to describe a KOS like its domain, language, and relations. It also models interactions between KOS through properties like has_version and is_interlinked_with. The goal is to enable discovery and understanding of existing KOS for improved reuse.
Multilingual Knowledge Organization Systems Management: Best PracticesMauro DragoniThe document discusses the modeling and management of multilingual Knowledge Organization Systems (KOS) in agriculture, highlighting their role in improving information access and reducing ambiguity. It outlines challenges such as multilinguality, collaborative work among experts, and the evolution of existing KOS, while introducing MOKI, a collaborative architecture using wikis and supporting multilingual ontology management. The Organic.Lingua project is mentioned, detailing its features, including connections to machine translation services and significant updates to the ontology during the project.
Wikipedia as source of collaboratively created Knowledge Organization SystemsJakob .The document discusses Wikipedia as a source of collaboratively created knowledge organization systems. It describes the structure of Wikipedia articles, categories, infoboxes, and how this structured data can be extracted and represented in semantic formats like RDF to create knowledge bases like DBpedia that link open data on the web. It also discusses some open issues around data quality, concepts and mapping when extracting and querying structured knowledge from Wikipedia.
Using OWL for the RESO Data DictionaryChimezie OgbujiThe document discusses the use of OWL (Ontology Web Language) for enhancing the RESO Data Dictionary to improve semantic interoperability and prevent misunderstandings in data communication. It outlines the structure of ontologies, including classes, properties, and relationships, and the benefits of using these for data representation in real estate contexts. Additionally, challenges and tools for implementing OWL in collaboration with domain experts are addressed, emphasizing the importance of well-designed ontologies for effective standardization.
DigiBird: on the fly collection integration using crowdsourcing VU University AmsterdamChris Dijkshoorn discusses challenges in crowdsourcing for data collection, including isolation, time demands, and integration into existing systems. The Digibird project proposes solutions like creating a hub for on-the-fly integration and using shared vocabularies to improve data access and collaboration. It emphasizes the need for continuous effort in promoting initiatives and refining contributions while acknowledging the complexities of developing a mature infrastructure.
Crowdsourcing Linked Data Quality AssessmentMaribel Acosta DeibeThe document presents a methodology for crowdsourcing the assessment of Linked Data quality. The methodology involves a two stage process - a find stage using Linked Data experts to identify potential quality issues, and a verify stage using microtasks on Amazon Mechanical Turk to validate the issues. The study assesses three types of quality problems in DBpedia through this methodology and analyzes the results in terms of precision. The findings indicate that crowdsourcing is effective for detecting certain quality issues, and that the expertise of Linked Data experts is best for domain-specific tasks while microtask workers perform well on data comparison tasks. The conclusions discuss integrating crowdsourcing into Linked Data curation processes and conducting further experiments.
Brief on Linked Data for U.S. EPA's Chief Data Scientist3 Round StonesThe document outlines the challenges and potential of data management at the EPA, highlighting the vast variety of data formats and silos that currently exist. It emphasizes the use of linked data and NoSQL platforms to improve data findability, accessibility, and interoperability, thus enhancing efficiency and transparency. Additionally, it discusses the implementation of a linked data management system to facilitate public access to environmental data and improve interactions across various datasets.
Role of Ontologies in Semantic Digital LibrariesSebastian Ryszard Kruk1) Ontologies play a key role in semantic digital libraries by supporting bibliographic descriptions, extensible resource structures, and community-aware features.
2) Semantic digital libraries integrate information from various metadata sources and provide interoperability between systems using semantics.
3) Key ontologies for digital libraries include bibliographic ontologies, structure description ontologies, and community-aware ontologies that model folksonomies and social semantic collaborative filtering.
Semantic Data Management in Graph Databases: ESWC 2014 TutorialMaribel Acosta DeibeThe document is a tutorial on semantic data management in graph databases, focusing on their structure and operations. It discusses graph data models, basic operations, various implementations, and the advantages of graph databases. Additionally, it covers existing graph database engines and their specific features for managing persistent and temporary graphs.
Linked Open Data in RomaniaVlad PoseaThis document discusses transforming open government data from Romania into linked open data. It begins with background on linked data and open data initiatives. Then it describes efforts to model, transform, link, and publish Romanian open data as linked open data. This includes identifying common vocabularies and properties, creating URIs, linking to external datasets like DBPedia, and publishing the linked data for use in applications via a SPARQL endpoint. Overall the goal is to make this data more accessible and interoperable through semantic web standards.
VIII Encuentros de Centros de Documentación de Arte Contemporáneo en Artium -...Artium VitoriaThe document outlines the development and vision for using linked data in cultural heritage on the semantic web from 2002 to 2016, emphasizing the creation of a collaborative framework for cultural institutions. It addresses the challenges of content complexity and production while presenting the 'sampo' model as a solution for semantic portals that facilitate access to rich cultural heritage resources. Three applications, including 'culturesampo' and 'warsampo', exemplify the effective use of linked data to enhance user engagement and research opportunities.
Wikidata and the Semantic Web of FoodBenjamin GoodThe document discusses the integration of Wikidata in building a semantic web of food knowledge, outlining its open and editable nature and its functionality as a knowledge base. It explains how Wikidata's structured items and statements create interconnected information, facilitating queries related to biomedical content, including genes, proteins, and diseases. The document also highlights various applications of this semantic web, supported by grants from the National Institute of Health.
Knowledge Organization System (KOS) for biodiversity information resources, G...Dag EndresenThe NCBO webinar focuses on the Global Biodiversity Information Facility (GBIF) and its work program aimed at enhancing the organization of biodiversity information. It emphasizes enabling free access to biodiversity data, and discusses the development of community standards, tools, and protocols for data management and publication. The presentation also outlines a governance structure for maintaining knowledge organization systems in biodiversity, along with recommendations for future developments in the field.
Knowledge Organization SystemsR A AkerkarThe document discusses knowledge organization systems (KOS) and how the Simple Knowledge Organization System (SKOS) bridges KOS and the Semantic Web. It provides examples of KOS like taxonomies and thesauruses and explains how they are used differently than ontologies. SKOS is defined as an RDF vocabulary for representing KOS online in a machine-readable way and became a W3C standard in 2009.
CHIST-ERA 2019 - presentation of CAMIL (Poznan University of Technology)Agnieszka ŁawrynowiczThe Center for Artificial Intelligence and Machine Learning (CAMIL) is a research organization in Poland that focuses on developing explainable machine learning systems and methodologies. Some of its areas of expertise include interpretable machine learning, decision support, active learning using explanations, and applying ontologies and knowledge graphs. It is interested in collaborating with others on projects involving using knowledge graphs and semantics for explainable AI.
Ontologie w historyczno-geograficznych systemach informacyjnychAgnieszka ŁawrynowiczPrezentacja z konferencji DARIAH-PL organizowanej przez PCSS. Wyniki analizy stanu badań w ramach projektu NPRH: Ontologiczne podstawy budowy historycznych systemów informacji geograficznej 2bH 15 0216 83
Semantic data mining: an ontology based approachAgnieszka ŁawrynowiczThe document discusses an ontology-based approach to semantic data mining, focusing on its integration with data mining methods and machine learning schemas. It outlines the use case of the e-lico intelligent discovery assistant and emphasizes the importance of background knowledge for optimizing data mining processes. The content includes a detailed explanation of the data mining optimization ontology (dmop) and its alignment with foundational ontologies for better representation of concepts and properties within data mining.
ML Schema: Machine Learning SchemaAgnieszka ŁawrynowiczThe document outlines the Machine Learning Schema (ML Schema) developed by Agnieszka Lawrynowicz and aims to create a shared framework for data mining and machine learning algorithms, datasets, and experiments in various formats. It seeks to align existing ontologies, support the integration of machine learning algorithms into linked open data, and promote collaboration with stakeholders like ML tool developers. Use cases include experiment sharing, meta-learning, and improving reproducibility in machine learning publications.
Semantic Meta-Mining of Knowledge Discovery ProcessesAgnieszka ŁawrynowiczThe document describes the e-LICO intelligent discovery assistant system. It consists of several components including a planner and meta-learner. The planner interacts with scientists to achieve their knowledge discovery goals through an iterative process. Other components include a data mining optimization ontology and services/components that deliver the data mining platform to scientists.
Hazardous Situation Ontology Design Pattern Agnieszka ŁawrynowiczThis document presents an ontology design pattern for modeling hazardous situations, outlining how objects are exposed to hazards and the consequences of such exposures. It includes formal definitions and competency questions to guide understanding of hazardous events. An example scenario illustrates the concepts, emphasizing the need for personal protective equipment to prevent health issues in occupational settings.
Using Substitutive Itemset Mining Framework for Finding Synonymous Properties...Agnieszka ŁawrynowiczThe document proposes a substitutive itemset mining framework to find synonymous properties in linked data. It applies frequent itemset mining to transactions of subject-property-object triples extracted from DBpedia to find property pairs that frequently co-occur. These pairs are then analyzed to identify substitutive properties that can replace each other based on their common coverage of itemsets. An implementation of this approach found several synonymous property mappings when tested on organization data from DBpedia.
Data Mining OPtimization Ontology and its application to meta-mining of knowl...Agnieszka ŁawrynowiczThis document describes the Data Mining Optimization Ontology (DMOP) and its application to meta-mining of knowledge discovery processes. DMOP aims to support decision making in data mining processes. It has about 750 classes, 200 properties, and 3200 axioms. The document discusses DMOP's modeling issues including meta-modeling, alignment with DOLCE foundational ontology, handling qualities and attributes, and use of property chains. It also describes applying DMOP to meta-learn from past data mining experiments in order to optimize future processes.
Data Mining OPtimization Ontology and its application to meta-mining of knowl...Agnieszka Ławrynowicz
Ad
ZTG 2013 Agnieszka Ławrynowicz
1. Wiedza w grach, gry z celem
tworzenia wiedzy
dr inż. Agnieszka Ławrynowicz
Instytut Informatyki Politechniki Poznańskiej
ZTG 2013
2. Kim jestem?
• Adiunkt w Instytucie Informatyki Politechniki
Poznańskiej
• Zainteresowania: sztuczna inteligencja,
głównie reprezentacja i inżynieria wiedzy
(ontologie), odkrywanie wiedzy i technologie
semantyczne (Semantic Web)
http://www.cs.put.poznan.pl/alawrynowicz/
3. LeoLOD
• LeoLOD - Learning and Evolving Ontologies from Linked
Open Data (2013-2015)
• Projekt realizowany w ramach programu POMOST Fundacji
na Rzecz Nauki Polskiej
• Tworzenie wiedzy: metody automatyczne (uczenie
maszynowe)
• Walidacja wyników: crowd-sourcing (mikro-zadania)
• Strona projektu:
http://www.cs.put.poznan.pl/alawrynowicz/leolod/
5. Jeopardy!
• Jeopardy! to amerykański quiz show (odpowiednik
polskiego Va Banque!)
• 1964 – do dzisiaj
• format odpowiedź-i-pytanie
• Przykład:
– Kategoria: Nauka ogólnie
– Wskazówka: W zderzeniu z elektronami, fosfor wydziela energię
elektromagnetyczną w tej formie
– Odpowiedź: Czym jest światło?
dla ludzi, wyzwaniem jest znajomość odpowiedzi
dla maszyn, wyzwaniem jest zrozumienie pytania
6. IBM Watson
• Watson – system komputerowy stworzony
przez IBM do odpowiadania na pytania
zadawane w języku naturalnym
• Watson wystąpił w Jeopardy! w trzydniowej
rozgrywce (2011) …
7. IBM Watson
…
• przeciwnikami IBM Watsona byli:
– Brad Rutter – do tej pory wygrał najwięcej
pieniędzy,
– Ken Jennings – był najdłużej niepokonanym
mistrzem
• IBM Watson zajął pierwsze miejsce
8. IBM Watson
• DeepQA (Watson)
– generuje i ocenia wiele hipotez wykorzystując kolekcję metod z
dziedziny przetwarzania języka naturalnego, uczenia
maszynowego, reprezentacji wiedzy i wnioskowania;
– gromadzą one i ważą dowody pochodzące ze źródeł danych
niestrukturalnych i strukturalnych (np. otwartych powiązanych
danych) aby ustalić odpowiedź o najwyższej pewności na
podstawie odpowiedzi wielu (setek) metod
JĘZYK NATURALNY ZADANIE
parsowanie
NER
wyszukiwanie
informacji
technologie
semantyczne
uczenie
maszynowe
crowd
9. IBM Watson
• DeepQA (Watson)
– generuje i ocenia wiele hipotez wykorzystując kolekcję metod z
dziedziny przetwarzania języka naturalnego, uczenia
maszynowego, reprezentacji wiedzy i wnioskowania;
– gromadzą one i ważą dowody pochodzące ze źródeł danych
niestrukturalnych i strukturalnych (np. otwartych powiązanych
danych) aby ustalić odpowiedź o najwyższej pewności na
podstawie odpowiedzi wielu (setek) metod
JĘZYK NATURALNY ZADANIE
parsowanie
NER
wyszukiwanie
informacji
technologie
semantyczne
uczenie
maszynowe
crowd
12. Tworzenie wiedzy
• wykwalifikowany zespół ludzi
• metody (pół)-automatyczne
• społecznościowe (crowd-sourcing):
– Gry z celem tworzenia wiedzy
13. Motywacje w tworzeniu treści przez
społeczność
• Obopólna korzyść (tagowanie)
• Reputacja, sława (Wikipedia)
• Rywalizacja
• Przystosowanie się do grupy
• Altruizm
• Poczucie własnej wartości i nauka
• Zabawa i osobista przyjemność
• Domniemana obietnica przyszłych nagród
• Nagrody (Amazon Mechanical Turk)
14. Gry z celem
• Games with a purpose (GWAP):
• Technika oparta na obliczeniach wykonywanych przez
ludzi (human-based computation)
• Proces obliczeniowy wykonywany jest poprzez zlecanie
niektórych czynności ludziom do wykonania w
zabawny, zajmujący sposób
• GWAP wykorzystuje różnice w umiejętnościach i
kosztach pracy ludzi i metod informatycznych w celu
osiągnięcia symbiotycznej interakcji człowiek-komputer
15. Gry z celem
• Luis Von Ahn (2006)
• Główna motywacja: nie leży w rozwiązaniu instancji
problemu obliczeniowego, jest to ludzkie pragnienie
zabawy
• W GWAP ludzie wykonują pożyteczne obliczenia jako
efekt uboczny przyjemnej rozrywki
• Miarą użyteczności GWAP jest kombinacja
wygenerowanych wyników i przyjemności rozgrywki
17. Gry z celem tworzenia treści, wiedzy
• Adnotacja tekstu/audio/obrazów/video
• Konstrukcja ontologii
• Mapowanie ontologii
• Tworzenie linków między zasobami
• „Wyścigi Wiki”
18. Adnotacja obrazów:
Google Image Labeler
• Dwuosobowa gra internetowa (online: 2006 – 2011,
wcześniej ESP Game)
• Cel: przypisanie etykiet do obrazka; dane wprowadzone
przez graczy wspomagały wyszukiwarkę grafik Google
• Zasady: punkty za podanie zgodnych etykiet obiektów na
obrazku. Często podawane etykiety umieszczane na
czarnej liście, niepunktowane.
• Dane wyjściowe : adnotacje opisujące obiekty na
obrazkach
• Walidacja: konsensus, większość
20. • Wieloosobowa gra
• Cel: adnotacja audio
• Zasady: kilka mini-gier dotyczących części utworu
muzycznego; wszyscy gracze słuchają tego
samego fragmentu audio i odpowiadają na
pytania. Punkty przyznawane za podobieństwo
odpowiedzi do tych udzielonych przez innych
graczy.
• Dane wyjściowe: adnotacja plików audio
• Walidacja: konsensus, większość
Adnotacja audio:
HerdIt
21. Ontologia w „pigułce”
• “An ontology is a
• formal specification maszynowa interpretacja
• of a shared grupa osób, konsensus
• conceptualization abstrakcyjny model zjawisk, pojęcia
• of a domain of interest“ wiedza dziedzinowa
• (Gruber 93)
ontologia = formalna specyfikacja pojęć z danej dziedziny
22. Konstrukcja ontologii:
OntoPronto (Ontogame)
• Dwuosobowa gra quizowa
• Cel: budowa ontologii dziedzinowej będącej
rozszerzeniem ontologii Proton
• Zasady: Gracze czytają streszczenie losowo
wybranego artykułu z Wikipedii i odpowiadają na
zapytania o relacji tego artykułu w stosunku do
ontologii Proton.
• Dane wyjściowe: Ontologia dziedzinowa
ufundowana na ontologii Proton
• Walidacja: konsensus, większość
24. Mapowanie ontologii:
SpotTheLink
• Dwuosobowa gra quizowa
• Cel: uzgadnianie ontologii, np. Dbpedia i Proton
• Zasady: Graczom prezentowane jest pojęcie z
jednej ontologii. Pierwszy krok: zgadzają się co do
odpowiadającego mu pojęcia w drugiej ontologii.
Krok drugi: zgadzają się co do relacji wiążącej te
dwa pojęcia.
• Dane wyjściowe: Odwzorowanie (w języku
SKOS)pomiędzy pojęciami w ontologiach
• Walidacja: konsensus, większość
26. Otwarte powiązane dane w „pigułce”
• Projekt społecznościowy ze wsparciem W3C
• Publikowanie zbiorów danych jako otwarte i powiązane ze
sobą dane grafowe (sieci semantyczne)
• Główna idea: wziąć istniejące (otwarte) zbiory danych i
uczynić je dostępnymi w sieci WWW w formacie RDF (sieci
semantyczne)
• Raz opublikowane w RDF, połączyć je linkami z innymi
zbiorami danych
• Przykładowy link RDF: http://dbpedia.org/resource/Berlin
[Identyfikator Berlina w DBPedia] owl:sameAs
http://sws.geonames.org/2950159 [Identyfikator Berlina w Geonames].
27. Tworzenie linków między zasobami:
VeriLinks
• Cel: walidacja linków w arbitralnym zbiorze
danych
• Zasady: Zgoda graczy co do poprawności linku
jest nagradzana monetami, które są następnie
wykorzystywane do zwalczania najeźdźców w
grze polegającej na obronie wieży.
• Dane wyjściowe: zwalidowane linki
29. „Wyścigi Wiki”:
Wikispeedia
• Podążanie za linkami w Wikipedii
• Cel: obliczanie semantycznej odległości
pomiędzy dwoma artykułami Wikipedii.
• Zasady: Gracze muszą znaleźć jak najkrótszą
ścieżkę między dwoma hasłami.
• Dane wyjściowe: semantyczna odległość
pomiędzy dwoma artykułami Walidacja:
Większość
31. Dalsze uwagi
• Nie każde zadanie da się łatwo przerobić na
GWAP (wymóg dekompozycji na mikro-
zadania)
• Tworzenie niektórych ontologii wymaga
bardzo specjalistycznej wiedzy
• To co powstaje w wyniku GWAP jest raczej
„płytkim” modelem
• GWAP wymaga strategii zapobiegania
oszustwom
32. Więcej informacji
• LeoLOD:
http://www.cs.put.poznan.pl/alawrynowicz/leolod
• IBM Watson (The DeepQA Project):
http://researcher.ibm.com/researcher/view_project.php?id=2099
• GWAP:
1. Luis von Ahn (2006). "Games With A Purpose" (PDF). IEEE Computer
Magazine: 96–98.
2. Luis von Ahn, Laura Dabbish (2008). "Designing Games With A Purpose"
(PDF). Communications of the ACM 51 (08/08).
• Semantic Games:
1. Elena Simperl, Roberta Cuel, Martin Stein, Incentive-Centric Semantic
Web Application Engineering, Morgan & Claypool Publishers (2013)
2. http://semanticgames.org/
Editor's Notes
#3: Może można od razu cos powiedziec o moim projekcie i ze tworzenie wiedzy jest jego tematem? Może można po prostu zrobic slajd z tego kim jestem i czym się zajmuje i drugi o kontekscie projektu i dlaczego mnie ten wlasnie temat/te tematy interesuje/interesuja?
#7: Można usunac np. ostatnia informacje i po tym puscic film z tego jak to bylo
#9: Tutaj może przed tym film a ten slajd jako podsumowanie
#10: Tutaj może przed tym film a ten slajd jako podsumowanie
#12: Tutaj może być krotka reklama, np., w pierwszym przypadku np., za pomoca narzedzia protege, w drugim chcemy napisac wtyczke do tego narzedzia i zrobic to pol-automatycznie, a dzisiaj chce mowic o trzecim temacie
#14: In the last few years researchers from different scientific disciplines investigated the grounds of the success of community-driven content creation. Although these studies reveal that the inner motivations that drive people to participate are heterogeneous and strongly context-specific, several main categories can be identified:
Reciprocity and expectancy Reciprocity means that contributors receive an immediate or long- term benefit in return for spending their time and resources on performing a certain task. An example of this is tagging, where the user organizes her knowledge assets, such as bookmarks or pictures, and while doing so, reuses tag classifications of other users.
Reputation Reputation is an important factor within (virtual) communities: it can drive users to invest time and effort to improve the opinion of their peers about them. It has shown that this is an important motivation for, for instance, Wikipedia users [75].
Competition Competition is a relevant incentive in the context of games (rankings), but it can also be a strong driver in community portals where the user with the, for example, most contributions, or highest status is awarded.
Conformitytoagroup Throughimitationandempathypeopletendtoconformtothesocialgroup they belong to, therefore making available information about members behaviors is a way to spur people to act according to this information. Staying with the example of Wikipedia as a strong community, studies have shown that the feeling of belonging to a group makes users be more active and supportive. They feel needed and thus obliged to contribute to the joint goals of their community [75].
Altruism People contribute to a joint endeavor because they believe it is for a good cause, without expecting anything in return.
Self-esteem and learning People contribute to a certain endeavor in order for them to grow as individuals, either in terms of their own self-perception or to increase their knowledge and skills.
Fun and personal enjoyment People get engaged in entertaining activities. The most popular ap- proach in this context hides a complex task behind casual games [114, 132].
Implicit promise of future monetary rewards People typically act in order to increase their own human capital and to position themselves to gain future rewards (money or better roles).
Rewards Peoplereceiveadirectcompensationwhichcanobviouslyplayalargeroleinexplainingthe rationale for contributing effort toward a goal. Examples of this are crowdsourcing platforms, such as Amazon’s Mechanical Turk2 or InnoCentive.3
#15: I tutaj o tym, ze wg tego pana najwazniejsza motywacja jest nie altruism i pieniadze ale wlasnie to radosc z grania
#17: described as a series of interconnected questions (or challenges), which the players will need to answer during the game.
The questions depend on the input, which is taken from the knowledge corpus the game processes and may be closed or open scale. The problem solved through the game needs to be highly decomposable into tasks which can be approached in a decentralized fashion by answering different questions. The questions themselves can have a varying level of difficulty, but in general it is assumed that an average player will be able to answer them in a matter of a couple of minutes.
The input used to generate the questions needs to be verbalized and translated into simple, unambiguous questions.
When the input is generated automatically, for instance because it contains results computed by algorithms that need to be validated by the players, it is essential that the quality of these results is not too low—otherwise the game experience will be less entertaining, as many of the questions will probably not make any sense at all.
One important aspect to be considered is how to assign questions to players. The basic approach is to do this randomly, though optimizations are possible when information about the performance of players in previous game rounds is available
In addition, in a multi-player game, the game developer might want to customize the ways players are matched to play against each other.
The output of a game with a purpose is an aggregated, manually cleansed form of the players’ answers. In a first step, the game developer needs to be able to match the contributions of the players; this is a not an issue for those challenges in which the set of allowed solutions are known, but it is less trivial in open-scale questions, where one has to deal not only with different spelling and spelling errors, but also synonyms and the like.
As a second step, the game developer needs ways to be able to identify correct answers with as little effort as possible, ideally automatically.
Once accurate answers are identified, the game developer needs to translate them into a semantic format, using existing ontologies and Semantic Web representation languages.
The most basic game elements are points and leader boards.
#18: Tutaj wymienione te kategorie od eleny lub z insemtives itp..