SlideShare une entreprise Scribd logo
1  sur  21
Télécharger pour lire hors ligne
Czech Technical 
University in 
Prague 
Personalised Access to Linked Data 
Milan Dojchinovski and Tomas Vitvar 
Web Intelligence Research Group 
Czech Technical University in Prague 
The 19th International Conference on Knowledge Engineering 
and Knowledge Management (EKAW 2014) 
November 24-28, 2014, Linköping, Sweden 
Milan Dojchinovski 
milan.dojchinovski@fit.cvut.cz - @m1ci - http://dojchinovski.mk 
Except where otherwise noted, the content of this presentation is licensed under 
Creative Commons Attribution-NonCommercial-ShareAlike 3.0 Unported 
Web Intelligence 
Research Group
Outline 
Personalised Access to Linked Data - @m1ci - http://dojchinovski.mk 
2 
• Introduction 
• Personalised Resource Recommendations 
• Experiments and Results 
• Conclusion and Future Work
Introduction 
Personalised Access to Linked Data - @m1ci - http://dojchinovski.mk 
3 
LOD cloud stats [1]: 
• 294 in Sep 2011 
• 1,091 datasets in Apr 2014 
• 271% growth 
• Find relevant information in LOD is not easy 
- SPARQL, manual dereferencing URIs, … 
• … or ask other people for recommendations and get 
personalised recommendations of resources 
• Linked Data based recommenders can help 
[1] M. Schmachtenberg et al, Adoption of linked data best practices in different topical domains, ISWC 2014.
Related Work 
Personalised Access to Linked Data - @m1ci - http://dojchinovski.mk 
4 
• dbRec (Passant, 2010): semantic distance measure 
- function of direct and indirect links 
• Content-based LD recommender (Di Noia et. al, 2012) 
- movies domain, max resource distance: 2 
• Lookup Explore Discovery (Mirizzi et al., 2010) 
- user input required 
- recommendations related to the entities occurring in the query 
• Discovery Hub (Marie et al., 2013) 
- based on the spreading activation 
- utilizes small portion of information DBpedia 
• Aemoo (Musetti et al., 2012) 
- Encyclopedic Knowledge Patterns over DBpedia
Introduction 
Personalised Access to Linked Data - @m1ci - http://dojchinovski.mk 
5 
• Method for personalised Linked Data recommendations 
- apply collaborative filtering technique to Linked Data 
- recommendations from users with similar resource interests 
• Two novel metrics: 
- resource similarity and resource relevance 
• Considered aspects: 
- Resource Commonalities 
- how much information two resources share 
- Resource Informativeness 
- how informative the resources are 
- Resource Connectivity 
- how well are resources connected
Outline 
Personalised Access to Linked Data - @m1ci - http://dojchinovski.mk 
6 
• Introduction 
• Personalised Resource Recommendations 
- Resource Similarity 
- Resource Relevance 
• Experiments and Results 
• Conclusion and Future Work
Resource Recommendation In a Nutshell 
ls:usedAPI 
#Hashtagram 
ls:usedAPI 
ls:usedAPI 
Personalised Access to Linked Data - @m1ci - http://dojchinovski.mk 
7 
• Input: RDF graph (including user profiles) 
• Step 1: evaluate user similarities 
- e.g. similarity between resources representing users 
- instances of foaf:Person class 
• Step 2: recommend resource from similar users 
- compute relevance for each resource candidate 
- incorporate the resource (user) similarities 
dc:creator 
dc:creator 
dc:creator 
creator dc:ls:category 
usedAPI 
ls:ls:usedAPI 
ls:tag 
ls:tag 
ls:tag 
ls:tag 
ls:usedAPI 
#microblogginig 
ls:tag 
ls:tag 
ls:tag 
#Alfredo 
#FriendLynx 
#Instagram 
#Twitter-API 
#Facebok-API 
#social 
#music 
#search #Microsoft-Bing- 
API 
#411Sync-API 
#MTV-Billboard-charts 
#Mobile- 
Weather-Search 
#mlachwani
Outline 
Personalised Access to Linked Data - @m1ci - http://dojchinovski.mk 
8 
• Introduction 
• Personalised Resource Recommendations 
- Resource Similarity 
- Resource Relevance 
• Experiments and Results 
• Conclusion and Future Work
Resource Similarity Computation 
dc:creator 
dc:creator 
ls:usedAPI 
#Hashtagram 
ls:usedAPI 
ls:usedAPI 
Personalised Access to Linked Data - @m1ci - http://dojchinovski.mk 
9 
• Assumption 1: the more information two resource share, 
the more similar they are 
#microblogginig 
ls:tag 
#social 
#music 
• 6 resources in the shared context graph 
dc:creator 
creator dc:ls:category 
usedAPI 
ls:ls:usedAPI 
ls:tag 
ls:tag 
ls:tag 
ls:tag 
ls:usedAPI 
ls:tag 
ls:tag 
#Alfredo 
#FriendLynx 
#Instagram 
#Twitter-API 
#Facebok-API 
#search #Microsoft-Bing- 
API 
#411Sync-API 
#MTV-Billboard-charts 
#Mobile- 
Weather-Search 
#mlachwani
Resource Similarity Computation (cont.) 
dc:creator 
#Instagram 
ls:tag 
#microblogginig 
ls:tag 
dc:creator 
ls:usedAPI 
#Hashtagram 
ls:usedAPI 
ls:usedAPI 
ls:usedAPI 
ls:tag 
ls:tag 
#Facebok-API 
#social 
#411Sync-API 
#music 
#Microsoft-Bing- 
ls:tag 
#Alfredo 
#FriendLynx 
#Twitter-API 
#search 
API 
Personalised Access to Linked Data - @m1ci - http://dojchinovski.mk 
Information Content (IC) 
10 
• Assumption 2: less probable shared resources carry more 
similarity information than the more common 
Resource IC 
#MTV-Billboard-charts 
dc:creator 
ls:tag 
ls:tag 
ls:usedAPI 
#mlachwani 
#Mobile- 
creator dc:Weather-Search 
ls:category 
usedAPI 
ls:• Evaluated by computing the node degree value 
- Microsoft-Bing-API (deg. 40) more than Twitter-API (deg. 799)
Resource Similarity Computation (cont.) 
dc:creator 
dc:creator 
ls:usedAPI 
#Hashtagram 
ls:usedAPI 
ls:usedAPI 
ls:usedAPI 
ls:tag 
#Alfredo 
#FriendLynx 
#Instagram 
#Twitter-API 
#microblogginig 
#social 
#411Sync-API 
#music 
#search #Microsoft-Bing- 
Personalised Access to Linked Data - @m1ci - http://dojchinovski.mk 
11 
• Assumption 3: better connected shared resources carry 
more similarity information 
#MTV-Billboard-charts 
dc:creator 
ls:tag 
ls:tag 
ls:usedAPI 
#mlachwani 
creator dc:Weather-Search 
ls:category 
usedAPI 
ls:ls:tag 
ls:tag 
ls:tag 
#Facebok-API 
ls:tag 
API 
#Mobile- 
• The number of simple paths between the resources 
- 2 simple paths between #Alfredo and #Twitter-API
Outline 
Personalised Access to Linked Data - @m1ci - http://dojchinovski.mk 
12 
• Introduction 
• Personalised Resource Recommendations 
- Resource Similarity 
- Resource Relevance 
• Experiments and Results 
• Conclusion and Future Work
Resource Relevance Computation 
dc:creator 
dc:creator 
ls:usedAPI 
#Hashtagram 
ls:usedAPI 
ls:usedAPI 
ls:usedAPI 
ls:tag 
#Alfredo 
#FriendLynx 
#Instagram 
#Twitter-API 
#mlachwani similar users 
#Facebok-API 
#microblogginig 
#social 
#411Sync-API 
#music 
#search #Microsoft-Bing- 
Personalised Access to Linked Data - @m1ci - http://dojchinovski.mk 
13 
• Recommending resources of type Web APIs for an user 
#MTV-Billboard-charts 
dc:creator 
ls:tag 
ls:tag 
ls:usedAPI 
creator dc:Weather-Search 
ls:category 
usedAPI 
ls:ls:tag 
ls:tag 
ls:tag 
ls:tag 
API 
#Mobile- 
• Recommendations from similar users 
- connectivity between the similar user and the resource candidate 
- number of simple paths 
- informativeness of each resource in these paths
Outline 
Personalised Access to Linked Data - @m1ci - http://dojchinovski.mk 
14 
• Introduction 
• Personalised Resource Recommendations 
- Resource Similarity 
- Resource Relevance 
• Experiments and Results 
• Conclusion and Future Work
Experiments Setup 
Personalised Access to Linked Data - @m1ci - http://dojchinovski.mk 
15 
• Linked Web APIs dataset 
- RDF representation of ProgrammableWeb.com 
- largest service and mashup repository 
• Evaluated accuracy and usefulness of recommendations 
• Accuracy: 
- precision/recall, AUC, NDCG, MAP, MRR 
• Usefulness: 
- serendipity: how surprising the recommendations are 
- diversity: how diverse the recommendations are 
• Evaluated methods: 
- User-KNN, Item-KNN, Most popular, Random 
- LD with RIC, LD without RIC
Accuracy Evaluation 
• Taking into account resource informativeness makes sense 
• Item-KNN and User-KNN do not work well 
- … at least in the Web services domain 
Personalised Access to Linked Data - @m1ci - http://dojchinovski.mk 
16 
0.0 0.2 0.4 0.6 0.8 1.0 
0.00 0.05 0.10 0.15 0.20 
Recall 
Precision 
Linked Data based with RIC 
Linked Data based without RIC 
User-KNN 
Item-KNN 
Most popular 
Random
Serendipity and Diversity Evaluation 
• Serendipity score = user resource avg. distance 
• Diversity score = avg. dissimilarity between all resource 
pairs 
Personalised Access to Linked Data - @m1ci - http://dojchinovski.mk 
17 
@top-N Random 
Most 
Popular 
User-KNN Item-KNN 
LD without 
RIC 
LD with 
RIC 
@top-5 2.97752 2.66810 2.59197 2.68006 3.18881 3.03271 
@top-10 2.98455 2.67465 2.65514 2.70402 3.54821 3.26700 
@top-15 2.98364 2.65816 2.68101 2.71267 3.73117 3.36509 
@top-20 2.98455 2.65184 2.69780 2.70968 3.84142 3.42444 
@top-5 0.65339 0.58347 0.62092 0.63349 0.83417 0.81949 
@top-10 0.65317 0.61354 0.62411 0.64392 0.86044 0.82912 
@top-15 0.65370 0.60374 0.63159 0.64558 0.87511 0.82884 
@top-20 0.65347 0.60719 0.63276 0.64287 0.88435 0.83114 
serendipity 
diversity
Trade-off: Serendipity, Diversity and Accuracy 
0.30 
0.28 
0.26 
0.24 
0.22 
0.20 
0.18 
0.16 
0.14 
0.12 
0.10 
0.08 
0.06 
0.04 
0.02 
@5 @10 @15 @20 
Personalised Access to Linked Data - @m1ci - http://dojchinovski.mk 
• higher serendipity leads 
to lower precision and 
higher recall 
• optimal results @top 5-10 
18 
0.30 
0.28 
0.26 
0.24 
0.22 
0.20 
0.18 
0.16 
0.14 
0.12 
0.10 
0.08 
0.06 
0.04 
0.02 
0.00 
Serendipity 
Precision 
0.834 
0.832 
0.830 
0.828 
0.826 
0.824 
0.822 
0.820 
0.818 
Diversity 
@5 @10 @15 @20 
0.73 0.74 0.75 0.76 0.77 0.78 0.79 0.80 0.81 0.82 
Recall 
Precision/Recall 
Diversity 
0.00 
Precision 
3.50 
3.45 
3.40 
3.35 
3.30 
3.25 
3.20 
3.15 
3.10 
3.05 
3.00 
0.73 0.74 0.75 0.76 0.77 0.78 0.79 0.80 0.81 0.82 
Recall 
Precision/Recall 
Serendipity 
• higher diversity leads to 
lower precision and 
higher recall 
• optimal results @top 5-10
Outline 
Personalised Access to Linked Data - @m1ci - http://dojchinovski.mk 
19 
• Introduction and Motivation 
• Personalised Resource Recommendations 
- Resource Similarity 
- Resource Relevance 
• Experiments and Results 
• Conclusion and Future Work
Conclusion 
Personalised Access to Linked Data - @m1ci - http://dojchinovski.mk 
20 
• Method for personalised access to Linked Data 
- recommendations based on the collaborative filtering 
technique 
• Considered aspects: 
- resources’ commonalities 
- resources’ informativeness 
- resources’ connectiviteness 
• Validated on a dataset from the Web services domain 
- Linked Web APIs dataset 
• Future work: 
- consider other multi-domain datasets 
- automatic determination of optimal resource contexts distances 
- publish the Linked Web APIs dataset to the LOD cloud
Feedback 
ls:usedAPI 
#Hashtagram 
ls:usedAPI 
ls:usedAPI 
Personalised Access to Linked Data - @m1ci - http://dojchinovski.mk 
21 
Thank you! 
Questions, comments, ideas? 
Milan Dojchinovski 
milan.dojchinovski@fit.cvut.cz 
@m1ci 
http://dojchinovski.mk 
dc:creator 
dc:creator 
dc:creator 
creator dc:ls:category 
usedAPI 
ls:ls:usedAPI 
ls:tag 
ls:tag 
ls:tag 
ls:tag 
ls:usedAPI 
#microblogginig 
ls:tag 
ls:tag 
ls:tag 
#Alfredo 
#FriendLynx 
#Instagram 
#Twitter-API 
#Facebok-API 
#social 
#music 
#search #Microsoft-Bing- 
API 
#411Sync-API 
#MTV-Billboard-charts 
#Mobile- 
Weather-Search 
#mlachwani

Contenu connexe

En vedette

Personalised Graph-Based Selection of Web APIs
Personalised Graph-Based Selection of Web APIsPersonalised Graph-Based Selection of Web APIs
Personalised Graph-Based Selection of Web APIsMilan Dojchinovski
 
Bodejuego
BodejuegoBodejuego
BodejuegoMar Mol
 
Finestra su europa
Finestra su europaFinestra su europa
Finestra su europahhartbree
 
Datasets and GATE Evaluation Framework for Benchmarking Wikipedia Based NER S...
Datasets and GATE Evaluation Framework for Benchmarking Wikipedia Based NER S...Datasets and GATE Evaluation Framework for Benchmarking Wikipedia Based NER S...
Datasets and GATE Evaluation Framework for Benchmarking Wikipedia Based NER S...Milan Dojchinovski
 
Chainable and Extendable Knowledge Integration Web Services: the FREME Framework
Chainable and Extendable Knowledge Integration Web Services: the FREME FrameworkChainable and Extendable Knowledge Integration Web Services: the FREME Framework
Chainable and Extendable Knowledge Integration Web Services: the FREME FrameworkMilan Dojchinovski
 
7 kalimah allah
7 kalimah allah7 kalimah allah
7 kalimah allahIcha Brow
 
Anggaran penjualan
Anggaran penjualanAnggaran penjualan
Anggaran penjualanIcha Brow
 
Presentase pemasaran
Presentase pemasaranPresentase pemasaran
Presentase pemasaranIcha Brow
 
01. kebijakan binsus (palopo) 2014
01. kebijakan binsus (palopo) 201401. kebijakan binsus (palopo) 2014
01. kebijakan binsus (palopo) 2014Icha Brow
 
Keuangan dan tata kelola lkp
Keuangan dan tata kelola lkpKeuangan dan tata kelola lkp
Keuangan dan tata kelola lkpIcha Brow
 
Prada H & D in Tokyo
Prada H & D in Tokyo Prada H & D in Tokyo
Prada H & D in Tokyo Emma Pereira
 
Manajemen mutu, visi, renstra
Manajemen mutu, visi, renstraManajemen mutu, visi, renstra
Manajemen mutu, visi, renstraIcha Brow
 

En vedette (15)

Personalised Graph-Based Selection of Web APIs
Personalised Graph-Based Selection of Web APIsPersonalised Graph-Based Selection of Web APIs
Personalised Graph-Based Selection of Web APIs
 
Bodejuego
BodejuegoBodejuego
Bodejuego
 
Finestra su europa
Finestra su europaFinestra su europa
Finestra su europa
 
Aress international
Aress internationalAress international
Aress international
 
Datasets and GATE Evaluation Framework for Benchmarking Wikipedia Based NER S...
Datasets and GATE Evaluation Framework for Benchmarking Wikipedia Based NER S...Datasets and GATE Evaluation Framework for Benchmarking Wikipedia Based NER S...
Datasets and GATE Evaluation Framework for Benchmarking Wikipedia Based NER S...
 
NIF Tutorial
NIF TutorialNIF Tutorial
NIF Tutorial
 
Chainable and Extendable Knowledge Integration Web Services: the FREME Framework
Chainable and Extendable Knowledge Integration Web Services: the FREME FrameworkChainable and Extendable Knowledge Integration Web Services: the FREME Framework
Chainable and Extendable Knowledge Integration Web Services: the FREME Framework
 
Humor kocak
Humor kocakHumor kocak
Humor kocak
 
7 kalimah allah
7 kalimah allah7 kalimah allah
7 kalimah allah
 
Anggaran penjualan
Anggaran penjualanAnggaran penjualan
Anggaran penjualan
 
Presentase pemasaran
Presentase pemasaranPresentase pemasaran
Presentase pemasaran
 
01. kebijakan binsus (palopo) 2014
01. kebijakan binsus (palopo) 201401. kebijakan binsus (palopo) 2014
01. kebijakan binsus (palopo) 2014
 
Keuangan dan tata kelola lkp
Keuangan dan tata kelola lkpKeuangan dan tata kelola lkp
Keuangan dan tata kelola lkp
 
Prada H & D in Tokyo
Prada H & D in Tokyo Prada H & D in Tokyo
Prada H & D in Tokyo
 
Manajemen mutu, visi, renstra
Manajemen mutu, visi, renstraManajemen mutu, visi, renstra
Manajemen mutu, visi, renstra
 

Similaire à Personalised Access to Linked Data

Linked Energy Data Generation
Linked Energy Data GenerationLinked Energy Data Generation
Linked Energy Data GenerationFilip Radulovic
 
Introduction to APIs and Linked Data
Introduction to APIs and Linked DataIntroduction to APIs and Linked Data
Introduction to APIs and Linked DataAdrian Stevenson
 
Linked Data at the OU - the story so far
Linked Data at the OU - the story so farLinked Data at the OU - the story so far
Linked Data at the OU - the story so farEnrico Daga
 
Walk Before You Run: Prerequisites to Linked Data
Walk Before You Run: Prerequisites to Linked DataWalk Before You Run: Prerequisites to Linked Data
Walk Before You Run: Prerequisites to Linked DataKenning Arlitsch
 
Developing rich interactive eBooks to teach linked open data to professionals...
Developing rich interactive eBooks to teach linked open data to professionals...Developing rich interactive eBooks to teach linked open data to professionals...
Developing rich interactive eBooks to teach linked open data to professionals...John Domingue
 
Linked services: Connecting services to the Web of Data
Linked services: Connecting services to the Web of DataLinked services: Connecting services to the Web of Data
Linked services: Connecting services to the Web of DataJohn Domingue
 
How To Structure Your Search Team for Success
How To Structure Your Search Team for SuccessHow To Structure Your Search Team for Success
How To Structure Your Search Team for SuccessOpenSource Connections
 
FAIR Dataverse
FAIR DataverseFAIR Dataverse
FAIR Dataversevty
 
FAIR principles and metrics for evaluation
FAIR principles and metrics for evaluationFAIR principles and metrics for evaluation
FAIR principles and metrics for evaluationMichel Dumontier
 
Search and Hyperlinking Overview @MediaEval2014
Search and Hyperlinking Overview @MediaEval2014Search and Hyperlinking Overview @MediaEval2014
Search and Hyperlinking Overview @MediaEval2014Maria Eskevich
 
#ALAAC15 Linked Data Love
#ALAAC15 Linked Data Love #ALAAC15 Linked Data Love
#ALAAC15 Linked Data Love Kristi Holmes
 
A Survey of Exploratory Search Systems Based on LOD Resources
A Survey of Exploratory Search Systems Based on LOD ResourcesA Survey of Exploratory Search Systems Based on LOD Resources
A Survey of Exploratory Search Systems Based on LOD ResourcesKarwan Jacksi
 
DataEngConf: Building Satori, a Hadoop toll for Data Extraction at LinkedIn
DataEngConf: Building Satori, a Hadoop toll for Data Extraction at LinkedInDataEngConf: Building Satori, a Hadoop toll for Data Extraction at LinkedIn
DataEngConf: Building Satori, a Hadoop toll for Data Extraction at LinkedInHakka Labs
 
Fox-Keynote-Now and Now of Data Publishing-nfdp13
Fox-Keynote-Now and Now of Data Publishing-nfdp13Fox-Keynote-Now and Now of Data Publishing-nfdp13
Fox-Keynote-Now and Now of Data Publishing-nfdp13DataDryad
 
Towards Semantic APIs for Research Data Services (Invited Talk)
Towards Semantic APIs for Research Data Services (Invited Talk)Towards Semantic APIs for Research Data Services (Invited Talk)
Towards Semantic APIs for Research Data Services (Invited Talk)Anna Fensel
 
Linked Data Workshop Stanford University
Linked Data Workshop Stanford University Linked Data Workshop Stanford University
Linked Data Workshop Stanford University Talis Consulting
 
Human Computation for Big Data
Human Computation for Big DataHuman Computation for Big Data
Human Computation for Big DataeXascale Infolab
 

Similaire à Personalised Access to Linked Data (20)

Linked Energy Data Generation
Linked Energy Data GenerationLinked Energy Data Generation
Linked Energy Data Generation
 
NISO Webinar: Library Linked Data: From Vision to Reality
NISO Webinar: Library Linked Data: From Vision to RealityNISO Webinar: Library Linked Data: From Vision to Reality
NISO Webinar: Library Linked Data: From Vision to Reality
 
Introduction to APIs and Linked Data
Introduction to APIs and Linked DataIntroduction to APIs and Linked Data
Introduction to APIs and Linked Data
 
Linked Data at the OU - the story so far
Linked Data at the OU - the story so farLinked Data at the OU - the story so far
Linked Data at the OU - the story so far
 
Walk Before You Run: Prerequisites to Linked Data
Walk Before You Run: Prerequisites to Linked DataWalk Before You Run: Prerequisites to Linked Data
Walk Before You Run: Prerequisites to Linked Data
 
Developing rich interactive eBooks to teach linked open data to professionals...
Developing rich interactive eBooks to teach linked open data to professionals...Developing rich interactive eBooks to teach linked open data to professionals...
Developing rich interactive eBooks to teach linked open data to professionals...
 
Linked services: Connecting services to the Web of Data
Linked services: Connecting services to the Web of DataLinked services: Connecting services to the Web of Data
Linked services: Connecting services to the Web of Data
 
Enabling Citizen-empowered Apps over Linked Data
Enabling Citizen-empowered Apps over Linked DataEnabling Citizen-empowered Apps over Linked Data
Enabling Citizen-empowered Apps over Linked Data
 
How To Structure Your Search Team for Success
How To Structure Your Search Team for SuccessHow To Structure Your Search Team for Success
How To Structure Your Search Team for Success
 
Alamw15 VIVO
Alamw15 VIVOAlamw15 VIVO
Alamw15 VIVO
 
FAIR Dataverse
FAIR DataverseFAIR Dataverse
FAIR Dataverse
 
FAIR principles and metrics for evaluation
FAIR principles and metrics for evaluationFAIR principles and metrics for evaluation
FAIR principles and metrics for evaluation
 
Search and Hyperlinking Overview @MediaEval2014
Search and Hyperlinking Overview @MediaEval2014Search and Hyperlinking Overview @MediaEval2014
Search and Hyperlinking Overview @MediaEval2014
 
#ALAAC15 Linked Data Love
#ALAAC15 Linked Data Love #ALAAC15 Linked Data Love
#ALAAC15 Linked Data Love
 
A Survey of Exploratory Search Systems Based on LOD Resources
A Survey of Exploratory Search Systems Based on LOD ResourcesA Survey of Exploratory Search Systems Based on LOD Resources
A Survey of Exploratory Search Systems Based on LOD Resources
 
DataEngConf: Building Satori, a Hadoop toll for Data Extraction at LinkedIn
DataEngConf: Building Satori, a Hadoop toll for Data Extraction at LinkedInDataEngConf: Building Satori, a Hadoop toll for Data Extraction at LinkedIn
DataEngConf: Building Satori, a Hadoop toll for Data Extraction at LinkedIn
 
Fox-Keynote-Now and Now of Data Publishing-nfdp13
Fox-Keynote-Now and Now of Data Publishing-nfdp13Fox-Keynote-Now and Now of Data Publishing-nfdp13
Fox-Keynote-Now and Now of Data Publishing-nfdp13
 
Towards Semantic APIs for Research Data Services (Invited Talk)
Towards Semantic APIs for Research Data Services (Invited Talk)Towards Semantic APIs for Research Data Services (Invited Talk)
Towards Semantic APIs for Research Data Services (Invited Talk)
 
Linked Data Workshop Stanford University
Linked Data Workshop Stanford University Linked Data Workshop Stanford University
Linked Data Workshop Stanford University
 
Human Computation for Big Data
Human Computation for Big DataHuman Computation for Big Data
Human Computation for Big Data
 

Dernier

New from BookNet Canada for 2024: BNC CataList - Tech Forum 2024
New from BookNet Canada for 2024: BNC CataList - Tech Forum 2024New from BookNet Canada for 2024: BNC CataList - Tech Forum 2024
New from BookNet Canada for 2024: BNC CataList - Tech Forum 2024BookNet Canada
 
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek SchlawackFwdays
 
Are Multi-Cloud and Serverless Good or Bad?
Are Multi-Cloud and Serverless Good or Bad?Are Multi-Cloud and Serverless Good or Bad?
Are Multi-Cloud and Serverless Good or Bad?Mattias Andersson
 
What's New in Teams Calling, Meetings and Devices March 2024
What's New in Teams Calling, Meetings and Devices March 2024What's New in Teams Calling, Meetings and Devices March 2024
What's New in Teams Calling, Meetings and Devices March 2024Stephanie Beckett
 
Unraveling Multimodality with Large Language Models.pdf
Unraveling Multimodality with Large Language Models.pdfUnraveling Multimodality with Large Language Models.pdf
Unraveling Multimodality with Large Language Models.pdfAlex Barbosa Coqueiro
 
Vertex AI Gemini Prompt Engineering Tips
Vertex AI Gemini Prompt Engineering TipsVertex AI Gemini Prompt Engineering Tips
Vertex AI Gemini Prompt Engineering TipsMiki Katsuragi
 
Leverage Zilliz Serverless - Up to 50X Saving for Your Vector Storage Cost
Leverage Zilliz Serverless - Up to 50X Saving for Your Vector Storage CostLeverage Zilliz Serverless - Up to 50X Saving for Your Vector Storage Cost
Leverage Zilliz Serverless - Up to 50X Saving for Your Vector Storage CostZilliz
 
The Future of Software Development - Devin AI Innovative Approach.pdf
The Future of Software Development - Devin AI Innovative Approach.pdfThe Future of Software Development - Devin AI Innovative Approach.pdf
The Future of Software Development - Devin AI Innovative Approach.pdfSeasiaInfotech2
 
Beyond Boundaries: Leveraging No-Code Solutions for Industry Innovation
Beyond Boundaries: Leveraging No-Code Solutions for Industry InnovationBeyond Boundaries: Leveraging No-Code Solutions for Industry Innovation
Beyond Boundaries: Leveraging No-Code Solutions for Industry InnovationSafe Software
 
Artificial intelligence in cctv survelliance.pptx
Artificial intelligence in cctv survelliance.pptxArtificial intelligence in cctv survelliance.pptx
Artificial intelligence in cctv survelliance.pptxhariprasad279825
 
CloudStudio User manual (basic edition):
CloudStudio User manual (basic edition):CloudStudio User manual (basic edition):
CloudStudio User manual (basic edition):comworks
 
Scanning the Internet for External Cloud Exposures via SSL Certs
Scanning the Internet for External Cloud Exposures via SSL CertsScanning the Internet for External Cloud Exposures via SSL Certs
Scanning the Internet for External Cloud Exposures via SSL CertsRizwan Syed
 
AI as an Interface for Commercial Buildings
AI as an Interface for Commercial BuildingsAI as an Interface for Commercial Buildings
AI as an Interface for Commercial BuildingsMemoori
 
Story boards and shot lists for my a level piece
Story boards and shot lists for my a level pieceStory boards and shot lists for my a level piece
Story boards and shot lists for my a level piececharlottematthew16
 
DevoxxFR 2024 Reproducible Builds with Apache Maven
DevoxxFR 2024 Reproducible Builds with Apache MavenDevoxxFR 2024 Reproducible Builds with Apache Maven
DevoxxFR 2024 Reproducible Builds with Apache MavenHervé Boutemy
 
WordPress Websites for Engineers: Elevate Your Brand
WordPress Websites for Engineers: Elevate Your BrandWordPress Websites for Engineers: Elevate Your Brand
WordPress Websites for Engineers: Elevate Your Brandgvaughan
 
Designing IA for AI - Information Architecture Conference 2024
Designing IA for AI - Information Architecture Conference 2024Designing IA for AI - Information Architecture Conference 2024
Designing IA for AI - Information Architecture Conference 2024Enterprise Knowledge
 
Bun (KitWorks Team Study 노별마루 발표 2024.4.22)
Bun (KitWorks Team Study 노별마루 발표 2024.4.22)Bun (KitWorks Team Study 노별마루 발표 2024.4.22)
Bun (KitWorks Team Study 노별마루 발표 2024.4.22)Wonjun Hwang
 
Powerpoint exploring the locations used in television show Time Clash
Powerpoint exploring the locations used in television show Time ClashPowerpoint exploring the locations used in television show Time Clash
Powerpoint exploring the locations used in television show Time Clashcharlottematthew16
 
Anypoint Exchange: It’s Not Just a Repo!
Anypoint Exchange: It’s Not Just a Repo!Anypoint Exchange: It’s Not Just a Repo!
Anypoint Exchange: It’s Not Just a Repo!Manik S Magar
 

Dernier (20)

New from BookNet Canada for 2024: BNC CataList - Tech Forum 2024
New from BookNet Canada for 2024: BNC CataList - Tech Forum 2024New from BookNet Canada for 2024: BNC CataList - Tech Forum 2024
New from BookNet Canada for 2024: BNC CataList - Tech Forum 2024
 
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack
 
Are Multi-Cloud and Serverless Good or Bad?
Are Multi-Cloud and Serverless Good or Bad?Are Multi-Cloud and Serverless Good or Bad?
Are Multi-Cloud and Serverless Good or Bad?
 
What's New in Teams Calling, Meetings and Devices March 2024
What's New in Teams Calling, Meetings and Devices March 2024What's New in Teams Calling, Meetings and Devices March 2024
What's New in Teams Calling, Meetings and Devices March 2024
 
Unraveling Multimodality with Large Language Models.pdf
Unraveling Multimodality with Large Language Models.pdfUnraveling Multimodality with Large Language Models.pdf
Unraveling Multimodality with Large Language Models.pdf
 
Vertex AI Gemini Prompt Engineering Tips
Vertex AI Gemini Prompt Engineering TipsVertex AI Gemini Prompt Engineering Tips
Vertex AI Gemini Prompt Engineering Tips
 
Leverage Zilliz Serverless - Up to 50X Saving for Your Vector Storage Cost
Leverage Zilliz Serverless - Up to 50X Saving for Your Vector Storage CostLeverage Zilliz Serverless - Up to 50X Saving for Your Vector Storage Cost
Leverage Zilliz Serverless - Up to 50X Saving for Your Vector Storage Cost
 
The Future of Software Development - Devin AI Innovative Approach.pdf
The Future of Software Development - Devin AI Innovative Approach.pdfThe Future of Software Development - Devin AI Innovative Approach.pdf
The Future of Software Development - Devin AI Innovative Approach.pdf
 
Beyond Boundaries: Leveraging No-Code Solutions for Industry Innovation
Beyond Boundaries: Leveraging No-Code Solutions for Industry InnovationBeyond Boundaries: Leveraging No-Code Solutions for Industry Innovation
Beyond Boundaries: Leveraging No-Code Solutions for Industry Innovation
 
Artificial intelligence in cctv survelliance.pptx
Artificial intelligence in cctv survelliance.pptxArtificial intelligence in cctv survelliance.pptx
Artificial intelligence in cctv survelliance.pptx
 
CloudStudio User manual (basic edition):
CloudStudio User manual (basic edition):CloudStudio User manual (basic edition):
CloudStudio User manual (basic edition):
 
Scanning the Internet for External Cloud Exposures via SSL Certs
Scanning the Internet for External Cloud Exposures via SSL CertsScanning the Internet for External Cloud Exposures via SSL Certs
Scanning the Internet for External Cloud Exposures via SSL Certs
 
AI as an Interface for Commercial Buildings
AI as an Interface for Commercial BuildingsAI as an Interface for Commercial Buildings
AI as an Interface for Commercial Buildings
 
Story boards and shot lists for my a level piece
Story boards and shot lists for my a level pieceStory boards and shot lists for my a level piece
Story boards and shot lists for my a level piece
 
DevoxxFR 2024 Reproducible Builds with Apache Maven
DevoxxFR 2024 Reproducible Builds with Apache MavenDevoxxFR 2024 Reproducible Builds with Apache Maven
DevoxxFR 2024 Reproducible Builds with Apache Maven
 
WordPress Websites for Engineers: Elevate Your Brand
WordPress Websites for Engineers: Elevate Your BrandWordPress Websites for Engineers: Elevate Your Brand
WordPress Websites for Engineers: Elevate Your Brand
 
Designing IA for AI - Information Architecture Conference 2024
Designing IA for AI - Information Architecture Conference 2024Designing IA for AI - Information Architecture Conference 2024
Designing IA for AI - Information Architecture Conference 2024
 
Bun (KitWorks Team Study 노별마루 발표 2024.4.22)
Bun (KitWorks Team Study 노별마루 발표 2024.4.22)Bun (KitWorks Team Study 노별마루 발표 2024.4.22)
Bun (KitWorks Team Study 노별마루 발표 2024.4.22)
 
Powerpoint exploring the locations used in television show Time Clash
Powerpoint exploring the locations used in television show Time ClashPowerpoint exploring the locations used in television show Time Clash
Powerpoint exploring the locations used in television show Time Clash
 
Anypoint Exchange: It’s Not Just a Repo!
Anypoint Exchange: It’s Not Just a Repo!Anypoint Exchange: It’s Not Just a Repo!
Anypoint Exchange: It’s Not Just a Repo!
 

Personalised Access to Linked Data

  • 1. Czech Technical University in Prague Personalised Access to Linked Data Milan Dojchinovski and Tomas Vitvar Web Intelligence Research Group Czech Technical University in Prague The 19th International Conference on Knowledge Engineering and Knowledge Management (EKAW 2014) November 24-28, 2014, Linköping, Sweden Milan Dojchinovski milan.dojchinovski@fit.cvut.cz - @m1ci - http://dojchinovski.mk Except where otherwise noted, the content of this presentation is licensed under Creative Commons Attribution-NonCommercial-ShareAlike 3.0 Unported Web Intelligence Research Group
  • 2. Outline Personalised Access to Linked Data - @m1ci - http://dojchinovski.mk 2 • Introduction • Personalised Resource Recommendations • Experiments and Results • Conclusion and Future Work
  • 3. Introduction Personalised Access to Linked Data - @m1ci - http://dojchinovski.mk 3 LOD cloud stats [1]: • 294 in Sep 2011 • 1,091 datasets in Apr 2014 • 271% growth • Find relevant information in LOD is not easy - SPARQL, manual dereferencing URIs, … • … or ask other people for recommendations and get personalised recommendations of resources • Linked Data based recommenders can help [1] M. Schmachtenberg et al, Adoption of linked data best practices in different topical domains, ISWC 2014.
  • 4. Related Work Personalised Access to Linked Data - @m1ci - http://dojchinovski.mk 4 • dbRec (Passant, 2010): semantic distance measure - function of direct and indirect links • Content-based LD recommender (Di Noia et. al, 2012) - movies domain, max resource distance: 2 • Lookup Explore Discovery (Mirizzi et al., 2010) - user input required - recommendations related to the entities occurring in the query • Discovery Hub (Marie et al., 2013) - based on the spreading activation - utilizes small portion of information DBpedia • Aemoo (Musetti et al., 2012) - Encyclopedic Knowledge Patterns over DBpedia
  • 5. Introduction Personalised Access to Linked Data - @m1ci - http://dojchinovski.mk 5 • Method for personalised Linked Data recommendations - apply collaborative filtering technique to Linked Data - recommendations from users with similar resource interests • Two novel metrics: - resource similarity and resource relevance • Considered aspects: - Resource Commonalities - how much information two resources share - Resource Informativeness - how informative the resources are - Resource Connectivity - how well are resources connected
  • 6. Outline Personalised Access to Linked Data - @m1ci - http://dojchinovski.mk 6 • Introduction • Personalised Resource Recommendations - Resource Similarity - Resource Relevance • Experiments and Results • Conclusion and Future Work
  • 7. Resource Recommendation In a Nutshell ls:usedAPI #Hashtagram ls:usedAPI ls:usedAPI Personalised Access to Linked Data - @m1ci - http://dojchinovski.mk 7 • Input: RDF graph (including user profiles) • Step 1: evaluate user similarities - e.g. similarity between resources representing users - instances of foaf:Person class • Step 2: recommend resource from similar users - compute relevance for each resource candidate - incorporate the resource (user) similarities dc:creator dc:creator dc:creator creator dc:ls:category usedAPI ls:ls:usedAPI ls:tag ls:tag ls:tag ls:tag ls:usedAPI #microblogginig ls:tag ls:tag ls:tag #Alfredo #FriendLynx #Instagram #Twitter-API #Facebok-API #social #music #search #Microsoft-Bing- API #411Sync-API #MTV-Billboard-charts #Mobile- Weather-Search #mlachwani
  • 8. Outline Personalised Access to Linked Data - @m1ci - http://dojchinovski.mk 8 • Introduction • Personalised Resource Recommendations - Resource Similarity - Resource Relevance • Experiments and Results • Conclusion and Future Work
  • 9. Resource Similarity Computation dc:creator dc:creator ls:usedAPI #Hashtagram ls:usedAPI ls:usedAPI Personalised Access to Linked Data - @m1ci - http://dojchinovski.mk 9 • Assumption 1: the more information two resource share, the more similar they are #microblogginig ls:tag #social #music • 6 resources in the shared context graph dc:creator creator dc:ls:category usedAPI ls:ls:usedAPI ls:tag ls:tag ls:tag ls:tag ls:usedAPI ls:tag ls:tag #Alfredo #FriendLynx #Instagram #Twitter-API #Facebok-API #search #Microsoft-Bing- API #411Sync-API #MTV-Billboard-charts #Mobile- Weather-Search #mlachwani
  • 10. Resource Similarity Computation (cont.) dc:creator #Instagram ls:tag #microblogginig ls:tag dc:creator ls:usedAPI #Hashtagram ls:usedAPI ls:usedAPI ls:usedAPI ls:tag ls:tag #Facebok-API #social #411Sync-API #music #Microsoft-Bing- ls:tag #Alfredo #FriendLynx #Twitter-API #search API Personalised Access to Linked Data - @m1ci - http://dojchinovski.mk Information Content (IC) 10 • Assumption 2: less probable shared resources carry more similarity information than the more common Resource IC #MTV-Billboard-charts dc:creator ls:tag ls:tag ls:usedAPI #mlachwani #Mobile- creator dc:Weather-Search ls:category usedAPI ls:• Evaluated by computing the node degree value - Microsoft-Bing-API (deg. 40) more than Twitter-API (deg. 799)
  • 11. Resource Similarity Computation (cont.) dc:creator dc:creator ls:usedAPI #Hashtagram ls:usedAPI ls:usedAPI ls:usedAPI ls:tag #Alfredo #FriendLynx #Instagram #Twitter-API #microblogginig #social #411Sync-API #music #search #Microsoft-Bing- Personalised Access to Linked Data - @m1ci - http://dojchinovski.mk 11 • Assumption 3: better connected shared resources carry more similarity information #MTV-Billboard-charts dc:creator ls:tag ls:tag ls:usedAPI #mlachwani creator dc:Weather-Search ls:category usedAPI ls:ls:tag ls:tag ls:tag #Facebok-API ls:tag API #Mobile- • The number of simple paths between the resources - 2 simple paths between #Alfredo and #Twitter-API
  • 12. Outline Personalised Access to Linked Data - @m1ci - http://dojchinovski.mk 12 • Introduction • Personalised Resource Recommendations - Resource Similarity - Resource Relevance • Experiments and Results • Conclusion and Future Work
  • 13. Resource Relevance Computation dc:creator dc:creator ls:usedAPI #Hashtagram ls:usedAPI ls:usedAPI ls:usedAPI ls:tag #Alfredo #FriendLynx #Instagram #Twitter-API #mlachwani similar users #Facebok-API #microblogginig #social #411Sync-API #music #search #Microsoft-Bing- Personalised Access to Linked Data - @m1ci - http://dojchinovski.mk 13 • Recommending resources of type Web APIs for an user #MTV-Billboard-charts dc:creator ls:tag ls:tag ls:usedAPI creator dc:Weather-Search ls:category usedAPI ls:ls:tag ls:tag ls:tag ls:tag API #Mobile- • Recommendations from similar users - connectivity between the similar user and the resource candidate - number of simple paths - informativeness of each resource in these paths
  • 14. Outline Personalised Access to Linked Data - @m1ci - http://dojchinovski.mk 14 • Introduction • Personalised Resource Recommendations - Resource Similarity - Resource Relevance • Experiments and Results • Conclusion and Future Work
  • 15. Experiments Setup Personalised Access to Linked Data - @m1ci - http://dojchinovski.mk 15 • Linked Web APIs dataset - RDF representation of ProgrammableWeb.com - largest service and mashup repository • Evaluated accuracy and usefulness of recommendations • Accuracy: - precision/recall, AUC, NDCG, MAP, MRR • Usefulness: - serendipity: how surprising the recommendations are - diversity: how diverse the recommendations are • Evaluated methods: - User-KNN, Item-KNN, Most popular, Random - LD with RIC, LD without RIC
  • 16. Accuracy Evaluation • Taking into account resource informativeness makes sense • Item-KNN and User-KNN do not work well - … at least in the Web services domain Personalised Access to Linked Data - @m1ci - http://dojchinovski.mk 16 0.0 0.2 0.4 0.6 0.8 1.0 0.00 0.05 0.10 0.15 0.20 Recall Precision Linked Data based with RIC Linked Data based without RIC User-KNN Item-KNN Most popular Random
  • 17. Serendipity and Diversity Evaluation • Serendipity score = user resource avg. distance • Diversity score = avg. dissimilarity between all resource pairs Personalised Access to Linked Data - @m1ci - http://dojchinovski.mk 17 @top-N Random Most Popular User-KNN Item-KNN LD without RIC LD with RIC @top-5 2.97752 2.66810 2.59197 2.68006 3.18881 3.03271 @top-10 2.98455 2.67465 2.65514 2.70402 3.54821 3.26700 @top-15 2.98364 2.65816 2.68101 2.71267 3.73117 3.36509 @top-20 2.98455 2.65184 2.69780 2.70968 3.84142 3.42444 @top-5 0.65339 0.58347 0.62092 0.63349 0.83417 0.81949 @top-10 0.65317 0.61354 0.62411 0.64392 0.86044 0.82912 @top-15 0.65370 0.60374 0.63159 0.64558 0.87511 0.82884 @top-20 0.65347 0.60719 0.63276 0.64287 0.88435 0.83114 serendipity diversity
  • 18. Trade-off: Serendipity, Diversity and Accuracy 0.30 0.28 0.26 0.24 0.22 0.20 0.18 0.16 0.14 0.12 0.10 0.08 0.06 0.04 0.02 @5 @10 @15 @20 Personalised Access to Linked Data - @m1ci - http://dojchinovski.mk • higher serendipity leads to lower precision and higher recall • optimal results @top 5-10 18 0.30 0.28 0.26 0.24 0.22 0.20 0.18 0.16 0.14 0.12 0.10 0.08 0.06 0.04 0.02 0.00 Serendipity Precision 0.834 0.832 0.830 0.828 0.826 0.824 0.822 0.820 0.818 Diversity @5 @10 @15 @20 0.73 0.74 0.75 0.76 0.77 0.78 0.79 0.80 0.81 0.82 Recall Precision/Recall Diversity 0.00 Precision 3.50 3.45 3.40 3.35 3.30 3.25 3.20 3.15 3.10 3.05 3.00 0.73 0.74 0.75 0.76 0.77 0.78 0.79 0.80 0.81 0.82 Recall Precision/Recall Serendipity • higher diversity leads to lower precision and higher recall • optimal results @top 5-10
  • 19. Outline Personalised Access to Linked Data - @m1ci - http://dojchinovski.mk 19 • Introduction and Motivation • Personalised Resource Recommendations - Resource Similarity - Resource Relevance • Experiments and Results • Conclusion and Future Work
  • 20. Conclusion Personalised Access to Linked Data - @m1ci - http://dojchinovski.mk 20 • Method for personalised access to Linked Data - recommendations based on the collaborative filtering technique • Considered aspects: - resources’ commonalities - resources’ informativeness - resources’ connectiviteness • Validated on a dataset from the Web services domain - Linked Web APIs dataset • Future work: - consider other multi-domain datasets - automatic determination of optimal resource contexts distances - publish the Linked Web APIs dataset to the LOD cloud
  • 21. Feedback ls:usedAPI #Hashtagram ls:usedAPI ls:usedAPI Personalised Access to Linked Data - @m1ci - http://dojchinovski.mk 21 Thank you! Questions, comments, ideas? Milan Dojchinovski milan.dojchinovski@fit.cvut.cz @m1ci http://dojchinovski.mk dc:creator dc:creator dc:creator creator dc:ls:category usedAPI ls:ls:usedAPI ls:tag ls:tag ls:tag ls:tag ls:usedAPI #microblogginig ls:tag ls:tag ls:tag #Alfredo #FriendLynx #Instagram #Twitter-API #Facebok-API #social #music #search #Microsoft-Bing- API #411Sync-API #MTV-Billboard-charts #Mobile- Weather-Search #mlachwani