Project Page | Documentation augmentation | High Level Deliverables | Detailed Deliverables | Meeting Notes | Hierarchy |
Year 3 Workplan
The following are the major component areas.
- Universal Hardware Platform (UHP)
- Detection Classification (DC)
- Data Management (DM)
- Project Integration
Final 6 Months of Year 3 Deliverables
This is the primary status document for Harder Project final 6 month deliverables.
Reference: (link to) Accepted Proposal for Final 6 Months
All items are listed in the order they are listed in the Proposal.
Item | Overall Progress | Priority | Lead | Major Deliverable (link to details for each) | Description | Supports Component Areas | Dependencies | Comments |
---|---|---|---|---|---|---|---|---|
1.0.0 | 3 | ELR | Interactive Workflow UI | Users will be able to interactively perform common steps for an analysis workflow to record metadata about sounds and events in the centralized database. Users will be able to interactively extract project metadata from the database. | Data Management, Project Integration | MQ Infrastructure, DM Internal Services | This work cannot begin until other pieces are in place. It is a lower priority than the Explore UI due to limited resources. | |
2.0.0 |
| 1 | ELR | Explore UI | Users will be able to navigate from projects to datasets to events to event. Users will be able to navigate through curated collections (see Curated Collections below). For this time frame, all queries will be limited to a single project or collection (see Filtered Queries in Beyond Year 3). | Data Management, Project Integration | MQ Infrastructure, DM Internal Services | On hold until dependencies are complete. |
3.0.0 |
| 1 | JR | Curated Collections | Users will be able to create a “collection” of sound events based on criteria specified by the curator. As a proof of concept, a curated NFC collection will be created using existing workflow XML configuration functionality to specify the criterion as a SQL Query. If time allows, the process for creating a curated collection will become available in the Workflow/Explore UI, removing the need for technical knowledge of XML and SQL Queries. | Data Management, Project Integration | ||
4.1.0 |
| 2 | ELR | Message Queue Integration (integration between software systems) | Users will be able to run a detector in MATLAB by launching it from Raven. Users will be able to view progressive results as they are generated by the detector in Raven Selection Review. Users will be able to export sound event metadata from the DM through the Explore UI. Users will be able to open a table in Raven from the Explore UI. If time allows, users will be able to run a detector through Delma by launching from Raven. All of these interactions between systems will be facilitated through Message Queue technology currently employed by DM to launch code running in different environments as part of the end-to-end workflow support. | Data Management, Project Integration | CRITICAL PATH | Includes...
|
4.2.0 |
| 1 | SK | Sandbox Database (integration between software systems) | Users will be able to create a local database on their machine for use during prototyping. Standard database interactions will be provided. Users will be able to create a table with standard column names and/or add user specified column names encouraging standards while allowing for flexibility. Users will be able to choose to export their data from the sandbox to the central DM. | Data Management, Detection Classification | MQ Infrastructure, DM Internal Services | Needs fetch() and publish() functionality. |
5.0.0 |
| 3 | MM | DM Scalability | Improve DM efficiency with a goal of ingesting a nd exporting millions of acoustic events in hours, not days. Long term, the use of a single-instance traditional relational database may impose limitations on efficiency for working at scale (see Distributed Database in Beyond Year 3). | Data Management | This was a top priority for the first half of FY14. Sufficient progress has been made to lower the priority. | |
6.0.0 | 1 | YS | Agnostic DC Approach | Explore the efficacy of a DC algorithm using unsupervised learning technique to produce a summary of types of signals in a sound stream. For example, a pulse and a Right Whale upsweep would not be grouped together, but a Right Whale upsweep and Humpback Whale upsweep would be grouped together. This will provide quick insight into the landscape of the sound stream allowing for a follow up approach that is targeted toward what is likely to be present. | Detection Classification |
First 6 Months of Year 3 Deliverables
End-to-End Scenarios
- (link to) Scenario 1
- (link to) Scenario 2
- (link to) Scenario 3
(link to) Suggestion Box - Ideas, Suggestions, Current Challenges