Trip Report ISWC 2016

October 31st, 2016

Wouter Beek

Which graph are we studying? (1)

W3C standards:

  • A set of triples (RDF Graph)
  • A named¹ set of triples (RDF Graph ≡ document contents)
  • Quads, i.e., a named² set of triples (RDF Graph ⊆ document contents)
  • ± entailment

Which graph are we studying? (2)

“Stream Reasoning Workshop”

  • A partial sequence of triples (RDF Stream)
  • A partial named¹ sequence of triples (RDF Stream from endpoint)
  • A partial sequence of quads, i.e., a named² sequence of triples
  • A named collection of named sets of statements (Quads + VoID)
  • ‘one-time semantics’ versus ‘continuous semantics’

Which graph are we studying? (3)

“Semantic Web meets Internet of Things and Web of Things”

  • The constantly changing(!) graph of interconnected devices
  • The graph of interconnected triple store endpoints
  • The graph of inter-linked HTML pages containing RDFa
  • The constantly changing and distributed collection of documents and endpoints that are referred to in the phrase “For our evaluation we have used Freebase.”

Implications for the study of meaning (1)

WoD: the intersection of meaning along the dimensions of model/set theory, sequent calculus, Web annotations, networking infra, and communication networks.

Meaning: formal, social, technological

What are ‘artifacts’ and what is ‘meaning’?

Implications for the study of meaning (2)

“Defining semantics of Linked Data on the WWW involves two graphs of different types: (1) an RDF graph distributed over documents and (2) the Web graph of documents.” [Hartig2015]

Hartig & Pirrò 2015, “A Context-Based Semantics for SPARQL Property Paths Over the Web”, ESWC

Baier & Daroch & Reutter & Vrgoč, 2016, “Property Paths over Linked Data: Can it be Done and How to Start?”, COLD.

Which graph are we studying? (5)

Mutiset Semantics of SPARQL Patterns

Multiset Relational Algebra (MRA) ≡ Mutliset Datalog (Mumuck et al. 1990)

“The Great Linked Data Debate”

  • Centralization (Ruben)
  • Open Data grows, but not RDF in Open Data (Axel)
  • Links are too expensive to maintain (Christian)
  • The Web is a platform for human communication.


Cochez & Decker & Prud'hommeaux, 2016, “Knowledge Representation on the Web Revisited: The Case for Prototypes”, ISWC.

Concept as network trace.

  • Simple change expression: $(p,\{r_1,\ldots,r_n\})$, with property $p$ and values $r_i$.
  • Prototype expression: $(id,(base,add,remove))$, with $add$ and $remove$ sets of simple change expressions, and with $base$ the prototype $id$ is derived from.

Combinatorial Creativity / Conceptual Blending

$Car \equiv Vehicle \sqcap \exists hasPart.Wheel$

A car without wheels is inconsistent.

Modification language: addition, subtraction, ‘succession’, selection, replacement.

Evaluation: Can existing inventions be described? Are the results of automatic manipulations considered meaningful?

Suchanek & Menard & Bienvenu & Chapellier, 2016, “Can You Imagine… A Language for Combinatorial Creativity?”, /ISWC/.

LOD Observatory (1)

2001 expectations:

  • People publish data on the Web
  • Ontologies are used to enable shared understanding
  • People implement cool & smart applications
  • Publishers turn into curators for quality

LOD Observatory (2)

Bizer keynote:

  • Most RDF datasets do not ‘link’
  • Lots of HTML-embedded (micro)data (e.g., >80% of PLDs in travel)
  • Cost/benefit of linking
  • “Making the Web a better place isn't enough motivation for most data providers.”

My take on this

  • Based on incomplete stats (Bizer2014)
  • There is way more LOD out there than the community is aware of (Rietveld2015).
  • Forgot 2 most important SW expectations: (1) agent/AI platform, (2) democratization platform
  • Linking ≠ ‘owl:sameAs’ (e.g., long/lat)

‘Grand Challenge’ (Hiroaki Kitano keynote)

“Scientific discovery is at pre-industry revolution level.”

“The engine of discovery should be a closed-loop system of hypothesis generation and verification, knowledge maintenance, knowledge integration, and so on.”

Thank you!