QUERYING HETEROGENEOUS LINKED BUILDING DATASETS WITH ...€¦ · •GraphQL (which one?)...

Post on 23-May-2020

2 views 0 download

Transcript of QUERYING HETEROGENEOUS LINKED BUILDING DATASETS WITH ...€¦ · •GraphQL (which one?)...

QUERYINGHETEROGENEOUS LINKED BUILDING DATASETSWITH CONTEXT-EXPANDED GRAPHQL QUERIES

JEROEN WERBROUCK

MADHUMITHA SENTHILVEL

JAKOB BEETZ

PIETER PAUWELS

LDAC2019-7th Linked Data in Architecture and Construction Workshop

• Renovation Projects

• BIM4REN

• Challenges

• ICDD

• Semantic Query Approaches• GraphQL-LD

• HyperGraphQL

• SparQL

• Comparison

• Implementation: Use Case

• Conclusions & Future Work

17.06.2019 LDAC2019 2

Outline

• Online framework for renovation-oriented BIM (H2020)

• Interconnecting heterogeneous information sources and tools represented as services

• Toolchains real-worldrenovation scenarios

• 23 partners accross Europe

• Practitioners, Developers, Researchers, …

17.06.2019 LDAC2019 3

BIM4REN

https://bim4ren.eu/

Heterogeneous sources in renovation context:

• Existing plans

• Point clouds

• Imagery

• IFC models

• Non-RDF databases

• Spreadsheets

• …

17.06.2019 LDAC2019 4

How to semantically relate this information?

LINKING EXISTING BUILDING SOURCES

• Information Container for Data Drop – ISO 21597(based on Dutch COINS standard)

17.06.2019 LDAC2019 3

ICDD

• Container OWL Ontology

• Linkset OWL ontology

Container-like Structure

• Describe document data with nodes

• Link them

Contains definitions

• Information linking on sub-document level

Terminologies

Example:

Spreadsheet cells

IFC GUIDs

Pixel/Voxel zones

Linked

• ZIP container

• Container OWL ontology

• Linkset OWL ontology

• File metadata in index.rdf

• Links in linkset.rdf

17.06.2019 LDAC2019 6

ICDD

ADVANTAGES:

• Bridges Linked Data research and industry“document-based” practice

• Ability to combine multiple exchange standards in a coherent way

17.06.2019 LDAC2019 8

ICDD - BALANCE

DRAWBACKS:

• No semantic classification of identifier types

• No implementation environments yet

• ICDD container as “dump file” of all project information

Alternative: Use ICDD ontologies for linkingproject source documents in a web-environment? (non-RDF + RDF)

• SPARQL

• Context-expanded GraphQL• HyperGraphQL (open source) (Semantic Integration Ltd.)

• GraphQL-LD (open source) (Ugent – IMEC)

• Stardog (commercial)

• TopBraid (commercial)

17.06.2019 LDAC2019 9

(Taelman et al., 2019)

SEMANTIC QUERY APPROACHES

when Domain experts are not semantic web experts

• Clean query writing

• Huge adoption rate

• Data and API querying

17.06.2019 LDAC2019 10

WHY GRAPHQL?

• Semantic integration Ltd.

• GraphQL-compliant• Context by GraphQL-schema + URIs and config.json

• “Lens” over Linked Data endpoint

• (data access regulated by schema)

• Endpoints:• SPARQL

• Local graphs

• Remote HyperGraphQL servers

• Intermediary service deployment

17.06.2019 LDAC2019 11

HYPERGRAPHQL

17.06.2019 LDAC2019 12

• UGent / IMEC

• GraphQL to SPARQL translation

• Context through JSON-LD

• Serverless

• Endpoints (via Comunica framework):• SPARQL

• Online RDF graphs (.ttl, .rdf, …)

17.06.2019 LDAC2019 13

GRAPHQL-LD

(Taelman et al., 2018)

17.06.2019 LDAC2019 14

GraphQL-LD HyperGraphQL SPARQL

Updating graphs No No Yes

Schema and Introspection No Yes (+ introspection) No

Intermediary server No Yes No

Federated service querying Yes Yes Yes

Reverse querying (O-P-S) Yes No Yes

17.06.2019 LDAC2019 15

COMPARISON

(Werbrouck et al., 2019b)

• Local copy of the project data

• Uploading ICDD graphs to triplestore

• (alternative HyperGraphQL: combine into single RDF graph)

• (alternative GraphQL-LD: store ICDD graphs online (cf next slide))

• Query

UC: LOCAL ICDD CONTAINERS

17.06.2019 LDAC2019 15

• Decentralised

• Many stakeholders in a project host their own data

• How to query the project data?• Non Linked Data-expert audience

• GraphQL (which one?)• HyperGraphQL

• needs new service setup when changing endpoint

• Concepts mapped to 1 single endpoint

• GraphQL-LD

• Embedded in application (Comunica or graphQL-to-SPARQL module)

• Flexible endpoint switching

• ICDD links between LBD/ifcOWL graphs and imagery, geometry, point clouds…

17.06.2019 LDAC2019 17

UC: LINKED DATA CDE

17.06.2019 LDAC2019 18

UC: LD-CDE QUERYING(Werbrouck et al., 2019a)

• GraphQL = Non-specialist way of query project data• Adding URI context for Linked Data use-cases

• Query documents on identifier (sub-document) level

• Local: GUI?

• Online: Web-applications• Potential for user-friendly LD-CDE

• Querying distributed linked building datasets

• Project sources (imagery, point clouds, geometry …) non-RDF

• Building Semantics RDF

• Communication mechanism between applications/APIs

17.06.2019 LDAC2019 19

CONCLUSION + FUTURE WORK

17.06.2019 LDAC2019 20

Thanks for your attention!

17.06.2019 LDAC2019 21

REFERENCES• Taelman, R., Vander Sande, M., Verborgh, R., 2019. Bridges between GraphQL and RDF.

Presented at the W3C Workshop on Web Standardization for Graph Data.

• Taelman, R., Vander Sande, M., Verborgh, R., 2018. GraphQLLD: Linked Data Querying with

GraphQL, in: ISWC2018, the 17th International Semantic Web Conference.

• Werbrouck, J., Pauwels, P., Beetz, J., van Berlo, L., 2019a. Towards a Decentralised Common

Data Environment using Linked Building Data and the Solid Ecosystem. Submitted for the

Proceedings of the 36th CIB W78 Conference on Information Technology in Construction.

• Werbrouck, J., Senthilvel, M., Beetz, J., Pauwels, P., 2019b. Semantic query languages for

knowledge-based web services in a construction context. To be presented at the 26th

International Workshop on Intelligent Computing in Engineering (EG-ICE), Leuven, Belgium.

17.06.2019 LDAC2019 7

ICDD – Link structure