Periodic Reporting for period 3 - ScienceGraph (Knowledge Graph based Representation, Augmentation and Exploration of Scholarly Communication)
Reporting period: 2022-05-01 to 2023-10-31
WP1 Deep Scholarly Knowledge Representation: We developed a conceptual model for representing scholarly contributions in knowledge graphs, which was published at the renowned JCDL 2020 conference . This work concluded MS1.2 Scholarly knowledge representation model and is currently implemented in MS1.3 Scalable graph storage.
WP2 Scholarly Knowledge Extraction, Graph Completion & Recommendation: We have successfully driven-forward work towards MS2.1 Knowledge graph-based information extraction by leveraging natural language processing methods for knowledge extraction from scientific literature resulting in a number of conference publications at LREC , ICADL and ECIR .
WP3 Knowledge-graph-based Communication & Interaction: We have devised and implemented work regarding MS3.1 Adaptive curation methods and published the results at JCDL with regard to the creation and curation of state-of-the-art comparisons and ICADL regarding importing and curating tables from survey articles.
WP4 Knowledge Graph Exploration & Question Answering has not yet formally started, but some preliminary work was done regarding simple question answering on state-of-the-art comparisons and published at TPDL .
WP5 ScienceGRAPH Testbed: We have started demonstrating the semantic organization of scholarly contributions leveraging knowledge graphs in the Open Research Knowledge Graph which we develop together with project partner TIB and is publically available at https://orkg.org. In particular, we have created state-of-the-art comparisons for more than 1.000 research problems (including regarding the COVID-19 pandemic), devised the concept of domain-specific observatories, and published first results in this regard.