Developing a machine learning algorithm
In the context of the CAPSELLA project we designed and develop a platform / infrastructure, which acts as the base layer of all the pilot applications being developed aiming on satisfying the basic needs of the farmers and the citizens in the related domains.
The CAPSELLA platform is designed under the following principles: (i) able to support different scenarios and requirements, (ii) to be extensible and pluggable, (iii) to implement the FAIR guidelines, (iv) to offer a set of independent services and (v) to provide data, metadata and analytics services agnostic of the various data types
To this end the current development activities of the platform have resulted in a number of services, listed below:
- The CAPSELLA Authentication service: The authentication service of CAPSELLA is exploited by the rest of the systems and by any application that is developed on top of the platform. It acts as a central point of authentication and authorisation and is based on the LDAP protocol
- The Data Management System: Offers data storage, retrieval and query facilities on top of the available datasets. It currently supports tabular data, geospatial, json and relational data types
- The Metadata Catalogue: The main CAPSELLA catalogue offering browse and search facilities over the available datasets based on their accompanied metadata. It is based on the well known CKAN data portal
- The CAPSELLA Data analytics and Management System: Offers analytics services, like sentiment analysis and opinion mining on various datasets, mostly focusing on social media sources, but also on the CAPSELLA generated datasets
- The CAPSELLA interoperability services: Contains a set of services for being interoperable with external OAI-PMH repositories and in addition to make available the CAPSELLA catalogue metadata using the OAI-PMH specification.
Develop a service that enables specific functionality and can be part of the CAPSELLA platform. Such a service could implement a specific machine learning algorithm (i.e. classification, prediction models) as a service which could be applied either on existing CAPSELLA datasets or new datasets which will then be published to the CAPSELLA Data Management System.
If you are interested, please subscribe to this challenge here.
Datasets to be used
A list of indicative CSV files that could be used for testing is available on: http://capsella.madgik.di.uoa.gr/dataset?res_format=CSV
Documentation is available on: https://ticketing.madgik.di.uoa.gr/redmine/projects/capsella-public/wiki/Platform_Services and all our REST services offer a Swagger interface.