KDDart – Knowledge Discovery System

KDDart

Visit http://www.kddart.org

The KDDart Knowledge Discovery System provides our clients with the ability to integrate three diverse types of data:

  • Phenotypic;
  • Genetic; and
  • Environmental.

Typically this system would be used for breeding and pre-breeding applications, however, KDDart has the flexibility of design to be applied to multidisciplinary agro-ecological and environmental research undertakings.

DArT initiated the development of the KDDart data integration platform working with industry clients over several years and it is now used by ‘early adopter’ clients and collaborators. To illustrate, data integration diagram 1 (click to enlarge) depicts how the central KDDart storage is fed from a variety of input mediums which in turn feeds data to perform other tasks such as:

  • Conducting analysis;
  • Managing inventories; and
  • Preparing for future trials/experiments.

Diagram 1 Click to enlarge

Clients benefit from a software platform built from the ground up on open IT industry standards consisting of a three layered architecture (Diagram 2), the middle layer of which is a RESTful Application Programming Interface (API) interface. This design helps to safeguard the time, resources and acquired knowledge in a system flexible enough to adjust to the rapid changes in technology and advancing breeding methods.

Diagram 2 Click to enlarge

The base layer provides effective and efficient storage services for a broad range of genotypic, phenotypic and environmental data. Central storage database(s) are only accessed via the middle Data Access Layer, or DAL, using the RESTful API. The DAL provides the most reliable, optimal and flexible mechanism for storing and consuming data for all users of KDDart.

The applications or top layer, consists of the IT tools for users to interact with the system. Using either applications that are DArT supplied or custom developed ‘in-house’, KDDart and the DAL will securely manage your data integrity and access.

To serve various types of breeding cycles several applications are available in the applications’ layer which are briefly introduced below. The following illustration of the cereal breeding cycle is a condensed example the roles each application is able to manage:

 

Diagram 3. Click to enlarge

Please Note: The KDDart infrastructure and applications are specifically designed for extensibility and flexibility to cater for a wide variety of scientific research and experimental activities, with as little restriction as possible. This introduction is very brief and only provides a few millimetres of insight into the depths KDDart’s capabilities. For further information please visit our www.kddart.org site.

  • medium_kdxploreBlock_0KDXplore is a versatile application, useful for breeders, technicians, curators and developers. Trial data now becomes accessible from many more angles.
    KDXplore can assist with your trial selection and manage the distribution of those copies onto multiple KDSmart devices. Once data has been collected in the field with KDSmart, it can move back to KDXplore for data curation then upload, to ensure only quality trial data is stored in KDDart.

    KDXplore provides assistance to users to enable working with data in either a graphical or tabular manner as the situation dictates.

    The KDXplore image at right provides a glimpse of the main trial navigator window showing trials, selected trial details and selected trial status with action buttons for the tasks that can be performed for that trial.

  • KDSmart logoKDSmart is an application designed to operate on a variety of Android tablet and phone devices for the collection of data in the field. Containing data selectively exported from KDDart, KDSmart captures and stores Trial measurement and result data in the field or nursery for subsequent uploading to KDDart.

    As needs dictate, KDSmart can operate in ‘standalone’ mode without the full KDDart infrastructure. Standalone can utilise trials defined in CSV files for import into the tool and export. So even without KDDart, KDSmart is still a very useful tool for capturing field and nursery data.

    The illustration at right shows one of the formats for the scoring window where most activity occurs. The user can select or hide traits for display and scoring at any time, along with a range of other options to suit the user or task at hand, even for convenient left or right hand scoring.

  • KDManage is a web browser application which does not require the installation of additional special software on user’s workstations. It enables users to perform a variety of administrative and day-to-day management tasks with their data stored in the KDDart database.

    A key tool from the initial configuration and preparation of your data in the KDDart database through to many day-to-day tasks such as data curation (see illustration). KDManage fulfills the administrative needs of adding data to satisfy entity dependencies and relationships which are not always hierarchical.

    As trials are defined and in progress, other activities within KDManage become more frequent. These may be importing trial results or the extraction, i.e. exporting, of data to conduct analysis of trial results using other software tools. Tools with direct access to KDDart via the Data Access Layer (DAL) can operate without the need for any data extraction. Analysis applications such as KDCompute operate in this manner.

  • small_kdsensBlockThe KDSens application provides an interface between various generic environmental sensors, such as weather stations, soil probes, etc. and the KDDart database using the Data Access Layer (DAL). Sensor definitions are maintained within KDDart.

    The KDSens program is designed to:

    • Collect data from single or multiple sensor stations (e.g. weather stations, soil probes, etc.);
    • Store the captured sensor data locally within files or database at the Sensor Station Manager;
    • Export stored data to the KDDart database using system’s API.

    Some of KDSens capabilities include;

    • Sensors can be added and viewed using a map;
    • Monitored and their data viewed graphically or in tabular form and analysed (as illustrated);
    • Command controls issued for the device; and
    • When max/min thresholds occur, actions or warnings configured to occur.
  • KDCompute is a web application and platform that provides capability to conduct efficient and extensible data analysis within a secure, customisable and cooperative plugin framework.

    Using your own or supplied algorithms, KDCompute can conduct analytical, imputation or data processing tasks efficiently leaving your computer free to perform other tasks.

    Installation of special software on your workstation/laptop/tablet is not needed; Only a web browser and network connection to your KDDart database. Tasks and workloads of multiple users are managed by the server component.

    Unless there is a specific need, data extraction is not required with KDCompute as data and analytics are integrated via the Data Access Layer API accessing your trial data directly in the KDDart repository.

Page last updated: August 17, 2018, 2:34 pm