Know-Center has successfully participated with a paper contributed to the Semantic Publishing Challenge, which was part of the ESCW15 conference

The European Semantic Web Conference (ESWC) is the most important conference for semantic web technologies in Europe. This year, the A-rated conference took place from 31st May until 4th June in Portoroz, Slovenia – with sunny weather and warm temperatures at the seaside it once more became a major venue for discussing the latest scientific results and technology innovations around semantic technologies.

The Know-Center contributed to ESWC2015 with a publication – this time it was actually a contribution to the Semantic Web Evaluation Challenge, specifically one of its subchallenges called “Semantic Publishing” (SemPub2015), which took place in the context of ESCW2015. One task of SemPub2015 consisted of extracting pieces of information such as (authors, affiliations, titles, or references) from scientific publications given in the PDF format and then providing the results in a linked dataset. The training set with PDF documents as well as the SPARQL-Queries to be answered were provided with the task description.

In this task, Stefan Klampfl and Roman Kern from the Knowledge Discovery Team at Know-Center successfully participated with their innovative technology for PDF extraction, which is the result of earlier research projects (CODE). The result can be seen in the submitted Know-Center Paper with the title “Machine Learning Techniques for Automatically Extracting Contextual Information from Scientific Publications”. This paper describes how one can extract information from scientific publications in PDF format by using supervised and unsupervised Machine Learning in a modular framework with trained classifiers and ontologies. Their approach provided to yield the best overall precision, third best recall value and third best F1 value (0.292) among all participants of the challenge (see summary of challenge results).

We congratulate Stefan Klampfl and Roman Kern to their excellent performance!

The paper is available for download here.

Complete bibliography:
S. Klampfl, R. Kern. Machine Learning Techniques for Automatically Extracting Contextual Information from Scientific Publications. In Semantic Web Evaluation Challenges 2015. CCIS. Springer, 2015.