Onotology-Based Annotation and Ranking Service for Geoscience
Abstract
There is a need to automatically annotate information using a either a control vocabulary or an ontology to make the information not only easily discoverable but also allow the information to be linked to other information based on these semantic annotations. We present an ontology annotation and a ranking service designed to address this need. The service can be configured to use an ontology describing a specific application domain. Given text inputs, this service generates annotations whenever the service finds terms that intersect both in the text and the ontology. The service is also capable of ranking the different inputs based on the "contextual" similarity to the information captured in the ontology. To rank a given input, the service uses a specialized algorithm which calculated both an ontological score based on precomputed weights of the intersecting term from the ontology and a statistical score using traditional term frequency- inverse document frequency (TF-IDF) approach. Both these scores are normalized and combined to generate the final ranking. An example application of this service to find relevant datasets for studying Hurricanes within NASA's data catalog. A hurricane ontology is used to index and rank all the data set descriptions from the metadata catalog and only the datasets that rank high are presented to the end users as contextually relevant for studying Hurricanes.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2012
- Bibcode:
- 2012AGUFMIN53D..05S
- Keywords:
-
- 1610 GLOBAL CHANGE / Atmosphere;
- 1958 INFORMATICS / Ontologies;
- 1968 INFORMATICS / Scientific reasoning/inference;
- 1970 INFORMATICS / Semantic web and semantic integration