EGOlink: Automatically Analyzing and Extending the Structure of Links in EGO
Affiliation: TU Darmstadt
External Partner: Leibniz-Institut für Europäische Geschichte, Mainz
Financial Support by the Funding Agency : Dec 2015 – Nov 2017
- Dr. Ines Grund, Leibniz-Institut für Europäische Geschichte
- Dr. Hatem Mousselly-Sergieh, TU Darmstadt
EGO (from German “Europäische Geschichte Online” – European History Online) is an open access portal for refereed academic articles (in English and German) on transcultural history of Europe. In contrast to traditional publications, EGO follows a Wikipedia-similar paradigm and links the articles to each other as well as to external resources, such as knowledge bases and multimedia objects. So far, these links were created manually by the editor of the articles. The aim of this pilot project is to reduce (redundant) human effort by developing (semi-)automatic methods for enriching EGO articles with inter- as well as external links and to assist the editors during the linking process.
The initial phase of EGOLink aims at investigating the current linking structures of EGO articles. Thereby, the manually created links as well as the co-citation patterns among the different articles will be statistically analyzed and visualized. The goal is to identify questionable structures, such as isolated or weakly linked articles. The next phase of EGOLink deals with the problem of automatically linking EGO articles to each other as well as to external resources like Wikipedia. The linking procedure puts a special emphasis on the evolutionary nature of the portal, i.e., establishing the links between already existing articles and new appearing ones (internal as well as external). Hereby, two main sub-problems will be addressed: 1) the automatic identification and ranking of potential anchors in EGO articles and 2) linking the identified anchors to their counterparts in other EGO articles and/or external resources. In the third phase of the project, the success of the presented solutions will be evaluated. The evaluation will be performed in form of expert user studies.
- Wolf, Konstantin: Entity Linking Annotation System for Historical Articles. Bachelor Thesis. Technische Universität Darmstadt, 2016.