Biomedical annotation metaserver
Integrating independent resources for biomedical text annotation visualisation, prediction and benchmarking biomedical entity recognition between independent automatic prediction systems.
Extended task deadlines. The final deadlines for the tasks are as follows:
CEMP/GPRO offline tasks:
- tests set release January 3rd 2017
- final submission January 15th 2017
TIPS online task:
- servers alive January 31st 2017
- server evaluation from February 1st till March 31st 2017
Biomedical types supported by servers@BeCalm
- Anatomic component
- Cell line and cell type
- Subcellular structure
- Tissue and organ
- Gene ontology
How It Works ?
A platform to visualise, predict and benchmark biomedical entity recognition between independent automatic prediction systems.
This platform interconnects multiple annotation servers with various recognition abilities. The aim is to offer users information of practical biomedical use.
The platform unites and standardised access to textual information extracted by various automatic systems interconnected worldwide.
Text miners may test the performance of their systems at the meta-server. The platform offers integrated access to high quality gold standards and state-of-the-art prediction systems.
Developed for text miners
Text miners can use the platform both for development purposes and to obtain annotations and classifications in a systematic and consistent way.
- Text miners can run periodic comparative analyses of their predictions against the annotation servers and the gold/silver standards available in the platform.
- System performance evaluation accounts for both recognition abilities and processing time. Precision, recall and F-score statistics are computed to understand and measure the relevance of the automatic annotations. The maximum and average latency times of the annotation servers are monitored continuously.
- The web service API enables developers to integrate the platform data with their own applications. For example, to combine the results obtained by multiple annotation pipelines.
Developed for end-users
The development of a platform that can operate freely on the complete set of Medline abstracts has the potential of being of great advantage to the biomedical community.
- Users may configure annotation projects according to their own purposes, including the set of documents to be processed, the annotation servers to be used and the biomedical annotations of interest.
- The platform provides detailed information on the skills of the various annotation servers, the exhibited performance, and current status.
- Results will be accessible via web browser and via web service API.
No installation required
The documents and annotations you want at any time and from anywhere. Without complications.
Periodic report on annotation server performance.
Web interface and web services API provide access to the platform. New automatic systems are always welcome.
Annotation server implementation is totally independent of the platform. Developers only have to implement the methods specified in the REST API.