Lim Ying Sean1, Arun Anand Sadanandan1, Dickson Lukose1 and

Klaus Tochtermann2

Scientific Publication

Retrieval in Linked Data

1 MIMOS Bhd, Kuala Lumpur, Malaysia, 2 Leibniz Information Centre for Economics (ZBW), Kiel, Germany

Overview

• Introduction

• Linked Datasets

• Prototype Implementation

• Future Work

• Conclusion

Introduction

• Linked Data allows libraries to create and deliver library

data that is sharable, extensible and easily re-usable.

• Through rich linkages with complementary data from

trusted sources, libraries can increase the value of their

library data beyond the sum of their sources taken

individually.

Introduction

• Some examples of Linked Data available.

4

Introduction

• Our goal is to identify and retrieve related

scientific publications from different Linked

Datasets published, from a single user interface.

• Scientific publications in Linked Data consist of 3

elements:

– Metadata

– Thesauri

– Name authority file

Linked Datasets

• Agrovoc

• OpenAgris

• Standard Thesaurus Wirtschaft (STW) - a thesaurus for

economics.

• EconStor - an open access server for free publication of academic

literature in economics.

Subset of OpenAGRIS Data Model

Thesaurus

Metadata

FOAF Vocabulary

Prototype Implementation

How does our prototype work?

(OpenAgris)

(EconStor) (STW)

Prototype Screenshot

Prototype Screenshot

OpenAgris EconStor

Prototype Screenshot

Future Work

• In order to improve the quality of retrieved

publications, there are some future

research works are required:

– Measure the relevancy of the related

publications.

– Enhance user experience when searching for

related publications.

Conclusion

• We have illustrated our work in consuming linked

datasets in the area of publications, and in particular we

described the process followed to retrieve related

publications from different linked datasets.

• The approach we adopted depends on thesaurus

alignment to retrieve related publications.

Pseudocode 1 Input: User input query Output: List of publication from local dataset Procedure: 1. Identify concepts from user query, Cn={C1,C2,….,Cn}; 2. Initialize the current concept pointer i; 3. Initialize an array of concepts Wn = {}; 4. while (i<n) do 5. if Ci has skos:narrower to another concept, S then 6. load concept S into the array, Wn= {S1}; 7. else 8. load concept Ci into the array, Wn = {S1,Ci…..}; 9. end if 10. Increase the current concept pointer i; 11. end while 12. Issues a SPARQL to local dataset to identify publications that consist of

concept Wn.

Pseudocode 2 Input: List of concept Output: List of publication from Linked Data Procedure: 1. Receive an array of concept An={C1,C2,…Cn}; 2. Initialize a 2 dimensions array S[src][j]; 3. Initialize the current concept pointer i; 4. while (i<n) do 5. if Ci has skos:exactMatch to another concept, SCONCEPT then 6. Identify the source for SCONCEPT, src; 7. S[src][j] SCONCEPT; 8. end if 9. Increase the current concept pointer i; 10. end while 11. Issues a SPARQL to Linked Data S[src] to identify publication that consist of

concept Sn, Sn∈ S[src][j].