Matchmaking Public Procurement Linked Open Data

• Published in: On the Move to Meaningful Internet Systems: OTM 2015 Conferences Vol. 9415; pp. 405 - 422
• Main Authors: Mynarz, Jindřich, Svátek, Vojtěch, Di Noia, Tommaso
• Format: Book Chapter
• Language: English
• Published: Switzerland Springer International Publishing AG 2015; Springer International Publishing
• Series: Lecture Notes in Computer Science
• Subjects: Artificial intelligence; Information retrieval; Linked open data; Matchmaking; Public procurement; Software Engineering; SPARQL
• ISBN: 9783319261478; 3319261479
• ISSN: 0302-9743; 1611-3349
• DOI: 10.1007/978-3-319-26148-5_27

An increasing amount of public procurement data is nowadays being ported to linked data format, in view of its exploitation by government, commercial as well as non-profit subjects. One of the crucial tasks in public procurement is matchmaking demand with supply. We conceived this task as that of finding a supplier with previous successful history of contracts similar to a current call for tenders. In this paper we show how to implement a portable matchmaking service that relies solely on the capability of SPARQL 1.1. In order to show its effectiveness, the proposed service has been tested and evaluated on the RDFized versions of 2 procurement databases: the European Union’s Tenders Electronic Daily and the Czech public procurement register. We evaluate several factors influencing matchmaking accuracy, including score aggregation and weighting, query expansion, contribution of additional features obtained from linked data, data quality and volume.