A virtual “Werkstatt” for digitization in the sciences- AD_INFRA1
- Microverse
- BExIS++
- BExIS
- Facilitating Semantic Data Interoperability and Integration for Citizen Science Sensor Data
- GFBio
- iBid
- Leveraging Knowledge Graphs for iDiv and Biodiversity
- INAS
- Jena Data Center (JDC)
- NFDI4Biodiversity
- Semantic Annotations for Building a Reproducible and Interoperable Solution for End-to-End Machine Learning Pipelines
- Semantic Description and (semi-) Automatic Annotation of Citizen Science Data
- ThurAI
- Wissensmodellierung und semantische Vernetzung heterogener Daten und Anwendungen
Semantic User Modeling for the Recommendation of Scientific Data
|
Startdate: 2016-01-01 Finishdate: 2017-12-31 Status: completed |
Member: Former Member: |
Description
Researchers in the medical and life sciences are faced with an increasing abundance of data in scientific repositories. Thus, handling and filtering of data becomes a more and more time-consuming and challenging task. Personalization techniques, in particular recommender systems, have been incorporated in current literature portals in order to assist users in finding the relevant information in an acceptable time. However, this support has not yet find its way in scientific data repositories, e.g., DRYAD or GBIF. Overall, conventional recommender systems need be improved since the recommendations are generated by matching user preferences against available data based on keywords and does not consider surrounding related terms.
Semantic recommender systems are a new rising research area that is using underlying knowledge bases, e.g., ontologies in the recommendation process. An ontology, a common vocabulary for sharing knowledge in a specific domain, allows a structured and meaningful storage of information in a certain context. Sharing and linking this knowledge can be achieved with the adherence of Linked Open Data (LOD) principles that guarantee a public access of information in a structured format with a concrete identifier and linked with other resources. Thus, semantic recommendations can present relevant results that go beyond a user’s stored preferences. One obstacle for semantic recommendation is the poor description of personalization settings in state-of-the-art semantic user model ontologies. Apart from storing general user interests it is not possible to specify further needs. Another problem with respect to scientific data is the insufficient opportunity to store a user’s publications or his or her academic background.
We aim at the establishment of a formal semantic user model for scientific research objects and its integration and usage in semantic recommender systems.