Towards Linked Data in Physics

Marcin Skulimowski

Faculty of Physics and Applied Informatics, University of Lodz, Pomorska 149/153, 90-236 Lodz, Poland

Keywords:

Semantic Web, Linked Data, Linked Science.

Abstract:

Linked Data refers to machine-understandable data published on the Web using the Resource Description

Framework (RDF). Publishing and linking data using RDF make data easier to discover and easier to use.

Nowadays, Linked Data principles have gained popularity in many ﬁelds. In this paper we are going to

examine the possibility of obtaining Linked Data in physics. In particular, we discuss the principles of Linked

Data in the context of publications on physics and present examples of Linked Data in a branch of physics

called quantum mechanics. Moreover we present a web tool supporting manual creation of Linked Data in

domains in which automating extraction of the data is not possible.

1 INTRODUCTION

Linked Data refers to machine-understandable data

published on the Web using the Resource Description

Framework (RDF). The key feature of RDF is that it

allows usto link identiﬁers for entities (e.g. digital ob-

jects, real objects, abstract concepts) and not just doc-

uments (as HTML). Moreover, RDF links are typed,

i.e. the nature of connection between two linked en-

tities can be stated explicitly. Publishing and link-

ing data using RDF will make data on the Web eas-

ier to discover, more accessible and, thus, easier to

use. Until now, Linked Data principles have gained

popularity in many ﬁelds and are applied to various

kinds of data, e.g. government data (Sheridan and

Tennison, 2010), commerce (Hepp, 2008), education

and social media (Heath and Bizer, 2011). Appli-

cations of Linked Data can be also found in various

branches of science, e.g. in geography (Sren Auer

and Hellmann, 2009), biomedicine (Ciccarese et al.,

2008), chemistry and biology (Wiljes and Cimiano,

2012). The Linked Open Data initiative of the Uni-

versity of M¨unster (LODUM)

, which aims to pub-

lish any non-sensitive data online according to the

Linked Data principles is worth mentioning (Kaup-

pinen et al., 2013).

The purpose of this study is to examine the possi-

bility of obtaining Linked Data in physics. Generally,

the data and knowledge are usually contained in re-

search articles and books. They usually involve con-

siderations using advanced mathematical formalism.

http://lodum.de

The data may also be formed into plots and tables

(e.g. data from experiments). These ways of storing

scientiﬁc data implies difﬁculties in data processing,

in particular searching for information. The search-

ing is usually restricted to full text search. Machines

are not able to integrate data from the articles. From

a machine point of view the articles are data islands.

And widely used citations are nothing more than rela-

tions between articles. As a consequence a statement

that article A cites article B has no precise meaning.

Does it mean that in article A the ideas and proposi-

tions from article B are creatively developed? Or does

it mean that in article A article B is only merely men-

tioned? To answer these questions we have to look

through the article. The question arises then whether

it is possible to represent data and knowledge avail-

able in physics in such a way that will enable inte-

gration, reusing and sharing the data i.e. according to

Linked Data principles. In order to answer this ques-

tion, we focus our attention on a branch of physics

called quantum mechanics. The main reason for this

choice is central importance of the theory for other

branches of physics and technology. Moreover, in or-

der to obtain Linked Data, we need formal represen-

tations of domain-speciﬁc terms which can be used in

RDF descriptions. In the case of quantum mechanics

we are developing an ontology which is suitable for

this purpose (Skulimowski, 2010).

In the rest of paper, we analyze the principles

of Linked Data in the context of data and knowl-

edge available in quantum mechanics. In particular,

we give an example of Linked Data for this disci-

Skulimowski M..

Towards Linked Data in Physics.

DOI: 10.5220/0004504700480054

In Proceedings of the 5th International Conference on Computer Supported Education (CSEDU-2013), pages 48-54

ISBN: 978-989-8565-53-2

 2013 SCITEPRESS (Science and Technology Publications, Lda.)

pline. We also present a draft version of a web tool

shortly called LYR (Link Your Research) which sup-

ports manual creation of Linked Data in domains in

which automated extraction of the data is not possi-

ble (e.g. quantum mechanics). The paper ends with a

short discussion and an outline of future work.

2 RELATED WORK

Previous work related to the presented research have

already been done. Several approaches for a creation

of linked data have been proposed. Among them there

are automated approaches and approaches supporting

manual creation (Wiljes and Cimiano, 2012). In order

to describe data appropriate vocabularies are needed.

As a result various ontologies have been developed,

e.g. supporting semantic publishing and referencing

(SPAR)

, scientiﬁc discourse

and open annotations

in science

. There also exists a number of systems

similar (in some sense) to LYR. ScienceWISE Plat-

form allows a community of scientists annotating and

bookmarking research articles using appropriate on-

tologies (Aberer et al., 2011). Annotations can be

also created using 4A framework based on idea ”an-

notations any where, annotations any time” (Smrz and

Dytrych, 2011). Scientiﬁc workﬂows and other arte-

facts can be published, shared and discovered using

Experiment Virtual Research Environment (Goble

and Roure, 2007).

3 LINKED DATA PRINCIPLES

AND QUANTUM MECHANICS

Tim Berners-Lee introduced Linked Data principles

describing a set of best practices for publishing struc-

tured data on the Web (Berners-Lee, 2006). Let us

now consider the principles in the context of data

and knowledge contained in research publications on

quantum mechanics.

3.1 Naming Things with URI

The ﬁrst Linked Data principle advocates naming en-

tities (real or abstract) with URIs (Heath and Bizer,

2011). However a question may arise, whether it is

possible for entities appearing in research articles on

quantum mechanics. In this case the entities usually

http://purl.org/spar/

http://purl.org/swan/1.2/discourse-relationships/

http://www.purl.org/ao/

correspond to some ’elements’ of mathematical struc-

ture of the theory e.g. observable, quantum state,

orthogonality. There are also entities corresponding

to the structure of an article e.g. deﬁnition, theorem,

lemma. In order to name these entities with a URI

we can utilize the Web accessibility of the research

article through the URL. And this URL can be used

to obtain URIs for the entities considered in the arti-

cle. Let us then consider a research article with the

following URI:

http://example.org/art3

We want to assign URIs to various entities from the

article. Assume, for example, that some Concept is

considered in the article. Then it is possible to name

this concept with the URI:

http://example.org/art3#Concept

It happens sometimes that an element considered in

an article (e.g. equation, condition) has no special

name. Inside of the article we can refer to it using

its number (e.g. ...equation (12)..., ...condition (7)...).

This number can be also used to assign a URI to the

element e.g.

http://example.org/art3#12

Standard elements of an article corresponding to its

structure (e.g. theorems, deﬁnitions, lemmas etc.) are

usually numbered too. These numbers can be also

used in naming the elements with URIs e.g.

http://example.org/art3#Theorem_3

http://example.org/art3#Definition_2

In addition, elements of the theory considered in arti-

cles are very often denoted by symbols e.g. ψ, H

. In

these cases it is possible to assign a URI using (sim-

pliﬁed) Latex symbols e.g.

http://example.org/art3#Psi

http://example.org/art3#H_2

for ψ and H

As the above examples show that it is possible to

name various entities from research publications with

URIs. Notice, however, that in the case when one en-

tity is named with two different URIs (e.g. using an

entity name and an entity number), URI aliases will

appear.

3.2 Providing Useful RDF Information

Descriptions of resources that are intended for ma-

chines should be represented in RDF. In general, it is

an impossible task to represent the whole scientiﬁc ar-

ticle on quantum mechanics using RDF triples. This

is mainly because of the complicated mathematical

TowardsLinkedDatainPhysics

formalism used in the theory which obviously can-

not be represented in RDF. We may, however, apply

a light-weight approach and concentrate only on RDF

links between elements of the theory considered in

some article and elements of the theory from other

articles or publications. Moreover, one may try to de-

termine types of the elements assigning them to vo-

cabulary terms (see Fig.1).

Figure 1: RDF links between entities from article A and

entities from articles B and C. The entities from article A

are also linked to classes from ontologies.

In order to create RDF links in quantum mechan-

ics we need an appropriate vocabularies. The vocab-

ularies are formally deﬁned in the following ontolo-

gies:

• quONTOm

- an OWL ontology describing main

concepts (e.g. observable, Hamiltonian, spin)

and relations (e.g. commutator, orthogonality) in

quantum mechanics. Aside from quantum me-

chanical concepts and relations, the ontology con-

tains elements which in fact do not belong to

the domain of quantum mechanics e.g. math-

ematical objects which should be contained in

an ontology for mathematics. Unfortunately, to

our best knowledge, an appropriate ontology for

mathematics, which could be imported and used

in quONTOm ontology does not exist. However,

we think that in the future these concepts will be

separated from the ontology and become parts of

auxiliary ontology (or ontologies).

• PHYSO (Physical Sciences Ontology)

- an on-

tology describing general concepts (e.g. princi-

ple, problem, assumption) and relations (e.g. has

property, consequence of) of physical sciences.

The ontology can be used to characterize compo-

nent parts and relations of any physical theory and

not only quantum mechanics.

• SACO (Scientiﬁc Article Content Ontology)

an ontology containing a set of objects proper-

ties enabling description of what is done/used

http://purl.org/quONTOm

http://purl.org/lyr/physo

http://purl.org/lyr/saco

in a research paper. In such article something

(e.g. some element of the theory) is analyzed, de-

scribed etc. Equivalently, using ’dummy’ subject,

we can say that the paper considers, describes

something (Glasman-Deal, 2010).

We stress that above ontologies are at the moment in-

complete and are gradually developed towards more

complete forms. Consequently, parts of them might

be changed in the near future.

3.3 Including Links to Other Things

The fourth principle of Linked Data recommends in-

cluding links to other URIs. These links are crucial

because they enable the discovery of additional data

resources. There are three types of RDF links.

Relationship Links. In one research article an en-

tity (e.g. equation, concept, deﬁnition) from another

article can be used (e.g. generalized). Assuming that

this element is named with a URI (according to one

of the methods considered above) we can create the

following RDF link pointing to this element:

art3:H phys:generalizes art6:H_0 .

A reference to some concept introduced in another ar-

ticle can be represented by the following link:

art3:Concept sac:isIntroducedIn

<http://ex2.org/art6> .

where

http://ex2.org/ar6

is a URI of the article.

Moreover, it is very often that some entity introduced

in one article is a solution to a problem considered in

another article. We can express this by the following

relationship link:

art3:10 phys:solutionTo art6:Theorem_5 .

Identity Links. An important part of RDF links in

Linked Data are identity links. An element of the the-

ory (e.g. operator, concept, deﬁnition) may be named

with two different symbols in two different articles.

In this case we can create the following identity link:

art3:V_1 owl:sameAs art6:W_1 .

It may happen that the same deﬁnition may have two

different numbers in different articles. We can repre-

sent the situation as follows:

art3:Def_1 owl:sameAs art6:Def_2 .

Identity links are very important because they enable

expression of different views on the same element of

the theory named with URI. Moreover, they enable

clients to retrieve further descriptions about the ele-

ment.

CSEDU2013-5thInternationalConferenceonComputerSupportedEducation

Vocabulary Links. The last type of RDF links are

vocabulary links pointing to deﬁnitions of vocabulary

terms. For example, in a research paper on quan-

tum mechanics some observable C

can be consid-

ered. Then we can create the following link:

art3:C_2 rdf:type quo:Observable.

It is shown in the next section that RDF links can be

used in RDF descriptions of research articles.

4 LINKING RESEARCH DATA

Let us assume now that we want to create RDF links

for some research paper on quantum mechanics. This

can be done in the following steps:

1. We choose entities from the article which we want

to describe in RDF (e.g. concept 7, formula (12),

etc.).

2. We name the entities with URIs (different ways

of creating URIs were presented in the previous

section).

3. Next we determine the relation between the paper

and the chosen elements (using e.g. SACO ontol-

ogy).

4. Using appropriate ontologies (e.g. PHYSO,

quONTOm), we determine types of chosen ele-

ments. In this way we create vocabulary links.

5. We add identity links pointing at URIs used in

other articles to identify the same entities.

6. We create relationship links between the entities

of the considered article and entities from other

articles.

We assume that RDF links obtained initially may

be incomplete. The quality and depth of detail can be

improved over time by adding new RDF links by the

research community. There are many reasons for this

’progressive enrichment’ (Dodds and Davis, 2011).

For example, the creator of the initial RDF links could

not know about some facts. Another reason may be

some change or improvement in the vocabulary that

was used. It very often happens that results of a pa-

per are used in some other paper published later in the

future. Then appropriate RDF links between the two

papers should be added.

4.1 Example

Let us now consider an example of Linked Data

for the case of quantum mechanics. To this

end we take into account a research paper avail-

able online at arXiv

with the following URL:

http://arXiv.org

http://http://arxiv.org/abs/1101.3969v1. The paper is

related to so called time operator problem. For con-

venience, we will use the following preﬁxes:

sac

Scientiﬁc Article Content Ontology,

quo

- quONTOm

Ontology,

- SWAN Discourse relationships vocab-

ulary.

In order to obtain RDF links we follow steps pre-

sented above:

1. We want to describe in RDF the following entities

from the article: M, g

m,λ

, σ(M).

2. URIs assigned to the entities:

<#M>

<#g_(m,lambda)>

<#sigma(M)>

3. Relations between the article and the entities:

<> sac:analyzes <#M> .

<> sac:determines <#g_(m,lambda)> .

<> sac:determines <#sigma(M)> .

4. Types of the entities:

<#M> rdf:type

quo:SelfAdjointOperator .

<#g_(m,lambda)> rdf:type

quo:EigenStates .

<#sigma(M)> rdf:type

quo:OperatorSpectrumBounded .

5. Identity link:

<#M> owl:sameAs

<http://link.aip.org/link/

doi/10.1063/1.3276419#M_F>

6. Relationship links:

<#M> quo:hasEigenStates

<#g_(m,lambda)> .

<#M> quo:hasSpectrum <#sigma(M)> .

<#M> sac:introducedIn

<http://link.aip.org/link/

doi/10.1063/1.3276419> .

<#M> dr:relatesTo

<http://arxiv.org/abs/

quant-ph/9611015#Theorem_1> .

Further examples of RDF abstracts for papers on

quantum mechanics can be found on the Link Your

Research project website

4.2 Beneﬁts

Let us now present probable beneﬁts of applying

Linked Data in quantum mechanic, and in physics

http://www.linkyourresearch.org/

TowardsLinkedDatainPhysics

in general. To this end assume that we have some

collection of scientiﬁc papers on quantum mechanics.

Today, only full text searching can be carried out on

the papers (eventually, with the help of keywords). In

the case when PACS (Physics and Astronomy Clas-

siﬁcation Scheme)

is used, searching for papers by

subject is also possible. Assume now that for each pa-

per RDF links are created and stored in a triplestore.

Then, using SPARQL endpoint one could make var-

ious queries against the dataset. For example, in the

case of RDF links created for papers related to time

operator problem in quantum mechanics the follow-

ing queries, among others, could be made:

• for some entities e.g. self-adjoint time operators

(

quo:SelfAdjointOperator

• for properties of a given entity e.g. for the type

of spectrum (

quo:Spectrum

) of some operator

(named with URI).

• for a relationship between given entities e.g. for

the commutator relation between some two oper-

ators (

quo:CommutatorRelation

• for entities with properties similar to a given en-

tity e.g. for time operators with a given spectrum

(

owl:sameAs

• for the article in which some entity was introduced

e.g. some time operator (

quo:introducedIn

• for articles in which some entity is analyzed e.g.

some time operator (

sac:analyzes

• for articles in which some entity is generalized

e.g. some quantum system (

sac:generalizes

It is obvious that possibility of asking such queries

opens new opportunities for the retrieval of informa-

tion in quantum mechanics. Scientiﬁc papers are no

longer only human-readable data islands. They be-

come in some part machine-understandable (at least

in some part). There are precise relationships between

papers and some entities described in them. Thanks to

that as results of queries we will obtain concrete en-

tities from scientiﬁc papers and not papers containing

some string (as today).

5 LYR WEB TOOL

Linked Data for research papers on quantum mechan-

ics has to be created manually by the research com-

munity, in particular by the author of a publication.

Automatic creation of valuable and precise relation-

ships between concepts and research papers seems to

http://publish.aps.org/PACS

be out of scope of this domain due to its complex-

ity. However, one can imagine tools supporting and

simplifying creation of such data. We have devel-

oped a prototype of such web tool called LYR (Link

Your Research)

. This tool supports creation of RDF

links for any kind of publication which is available

online. Such a publication is called in LYR a con-

text. Each context has its URI identiﬁer (usually it is a

URL of the publication). A context corresponding to

some publication contains RDF links created by the

research community. In order to use the tool one has

to register and possess a personal URI identiﬁer. Af-

ter simple registration one can create a new context or

add links to already created contexts. The process of

adding links has a few steps (including six steps de-

scribed in Section 4) and consists of ﬁlling up a sim-

ple form available on the website. To this end one

has to know terms from appropriate ontologies (see

Fig.2). A user can choose ontologies he wants to use

from available list. At present, there are only a few

ontologies available however, new ontologies will be

successively added (any user can suggest adding new

ontology).

All RDF links generated using the LYR tool are

stored in Virtuoso Open-Source RDF Triple Store

The dataset obtained so far consists of a few hundred

of links (RDF triples) generated for several dozen of

research articles. Each article is available on the jour-

nal’s website or at arXiv. Most of the articles are re-

lated to time operator problem in quantum mechanics.

The reason for this limitation is the content of the on-

tology for quantum mechanics. In the actual version

of the ontology only the most fundamental concepts

and relations of the theory are represented, and the

time operator problem is related to such fundamental

concepts. The test dataset is gradually enlarged to in-

clude more links corresponding to various articles.

Figure 2: LYR - adding links to a context.

The tool will be available soon for tests. Please

visit our project website for more infornation at

http://www.linkyourresearch.org

http://virtuoso.openlinksw.com/

CSEDU2013-5thInternationalConferenceonComputerSupportedEducation

Figure 3: LYR - search for a resource.

Figure 4: LYR - search for a term.

The LYR tool enables easy search and exploration of

the dataset. One can search the dataset for all RDF

links from some context, an entity (resource) from

some publication (see Fig.3) or a term. In the last

case as a result one obtains resources ”corresponding”

to that term and not strings containing the term (see

Fig.4). It is also possible to retrieve links between re-

sources using SPARQL. Finally, it is worth to notice

that requesting the linked data corresponding to some

context is also possible. To this end one has to use the

following URL:

http://www.linkyourresearch.org/contextURI

where

contextURI

is the URI of a requested context.

When using this URL in a browser a web page con-

taining all RDF triples from the context is loaded. In

order to facilitate the creation of RDF links using the

LYR tool a Firefox browser extension is developed.

The extension will allow to see and add RDF triples

for a visited research article (context).

6 DISCUSSION AND FUTURE

WORK

In this preliminary paper we consider the Linked Data

in the context of physics. In particular, we exam-

ine the possibility of representing physics data and

knowledge according to the principles. We focus on

research articles on quantum mechanics and showthat

it is possible to obtain Linked Data in those cases, us-

ing terms from appropriate ontologies. In particular,

it is possible to link elements of the theory from two

papers, not just documents containing these papers.

There are, however, two important issues deserving

attention. First, we want to stress that the obtained

RDF links correspond only to part of human-readable

data available in the articles. There are two reasons

of the fact. Lack of the appropriate vocabularies -

presented ontologies remain still under development.

Another reason is that we are not able to represent ad-

vanced mathematical structures of the theory in RDF

and OWL (Skulimowski, 2010). We also want to no-

tice that automatic creation of RDF links for scientiﬁc

publications in physics seems to be out of scope in

this domain. The links has to be created manually by

the research community. Taking this into account we

are developing a web tool which support the creation.

The tool is shortly presented in the paper.

Future work should focus on development of the

presented ontologies. In particular, ’harmonization’

between the ontologies and others ontologies (e.g. Se-

mantic Publishing and Referencing (SPAR) ontolo-

gies and SWAN Scientiﬁc Discourse Ontology may

be required to remove overlap and conﬂicts. More-

over, it would be beneﬁcial to provide general hints

on how to create Linked Data for the case of research

articles on physics. Thanks to such hints creation of

Linked Data for a research articles could become a

part of scientiﬁc activity and even a part of an edito-

rial process in the future. The future development of

the LYR tool should focus on increasing support dur-

ing the creation of RDF links, for example automatic

generation of URIs or suggesting links between re-

sources. Moreover, visualization possibilities of RDF

links stored in LYR should be extended. In future,

care should be also taken to ensure consistency of

RDF data created using the tool.

REFERENCES

Aberer, K., Boyarsky, A., Cudr´e-Mauroux, P., Demartini,

G., and Ruchayskiy, O. (2011). Sciencewise: A web-

based interactive semantic platform for scientiﬁc col-

laboration. In Proceedings on the 10th International

TowardsLinkedDatainPhysics

Semantic Web Conference (ISWC2011), Bonn, Ger-

many.

Berners-Lee, T. (2006). Linked data - design issues. Re-

trieved January 12, 2013, from http://www.w3.org/

DesignIssues/LinkedData.html.

Ciccarese, P., Wub, E., Wongb, G., Ocana, M., Kinoshita,

J., Ruttenberg, A., and Clark, T. (2008). The SWAN

biomedical discourse ontology. Journal of Biomedical

Informatics, 41(5):739–751.

Dodds, L. and Davis, I. (2011). Linked Data Patterns:

A pattern catalogue for modelling, publishing, and

consuming Linked Data. Retrieved February 7, 2013,

from http://patterns.dataincubator.org/book/linked-

data-patterns.pdf.

Glasman-Deal, H. (2010). A Guide for Non-Native Speak-

ers of English. London: Imperial College Press.

Goble, C. and Roure, D. D. (2007). myExperiment: So-

cial networking for workﬂow-using e-scientists. In

Proceedings of the 2nd workshop on Workﬂows in

support of large-scale science, Monterey, California,

USA, ACM, pages 1–2.

Heath, T. and Bizer, C. (2011). Linked Data: Evolving the

Web into a Global Data Space. Morgan & Claypool

Publishers.

Hepp, M. (2008). Goodrelations: An ontology for de-

scribing products and services offers on the web.

In Proceedings of the 16th International Conference

on Knowledge Engineering and Knowledge Manage-

ment, Acitrezza, Italy.

Kauppinen, T., Baglatzi, A., and Keler, C. (2013). Linked

science: Interconnecting scientiﬁc assets. In Terence

Critchlow and Kerstin Kleese-Van Dam (Eds.): Data

Intensive Science. CRC Press, USA.

Sheridan, J. and Tennison, J. (2010). Linking UK govern-

ment data. In Proceedings of the Linked Data on the

Web Workshop (LDOW2010).

Skulimowski, M. (2010). An OWL ontology for quan-

tum mechanics. In Proceedings of the 7th Inter-

national Workshop on OWL: Experiences and Di-

rections (OWLED 2010), San Francisco. Retrieved

February 18, 2013, http://sunsite.informatik.rwth-

aachen.de/Publications/CEUR-WS/Vol-614/.

Smrz, P. and Dytrych, J. (2011). Towards new scholarly

communication: A case study of the 4A framework.

In Proceedings of the Workshop on the Semantic Pub-

lishing (SePublica). Retrieved February 18, 2013,

http://ceur-ws.org/Vol-721/paper-07.pdf.

Sren Auer, J. L. and Hellmann, S. (2009). Linkedgeodata -

adding a spatial dimension to the web of data. In Pro-

ceedings of International Semantic Web Conference.

Wiljes, C. and Cimiano, P. (2012). Linked data for the nat-

ural sciences: Two use cases in chemistry and biol-

ogy. In Proceedings of the Workshop on the Semantic

Publishing (SePublica). Retrieved February 18, 2013,

http://ceur-ws.org/Vol-903/paper-07.pdf.

CSEDU2013-5thInternationalConferenceonComputerSupportedEducation