Skip to content

Latest commit

 

History

History
71 lines (44 loc) · 4.25 KB

roadmap.md

File metadata and controls

71 lines (44 loc) · 4.25 KB

I3 Roadmap

This section details the development process for the I3 project. It is understood that the I3 proof-of-concept is likely to take many iterations to perfect with the refinements to the manifest, worker layout and other tooling to change along the way.

Creation of CLI proof-of-concept (complete)

The first step for the I3 project will be to create a command line interface (CLI) program which implements the base standard. Since CLI applications are easier to develop and debug this should test the manifest specification against IEBH's existing modules before promoting the standard to a wider audience.

This initial stage will include the task of creating the base project, implementing the manifest schema and testing against a selection of simple workers.

Scenario testing (in progress)

The next stage is using the CLI tool to run though a set scenario and testing the viability at each stage.

A proposed scenario would be as follows:

# Scenario Tool Test purpose Input Output
1 Deduplication IEBH-Dedupe Command line streaming as a block library EndNote XML Deduplicated library, duplicates
2 Spider references by one generation forward / back IEBH-Spider Command line streaming individual references EndNote XML Original + Spidered references
3 Screen the references Abstrackr Support for legacy, self hosted tools CSV Partial CSV data

Notes:

  • Abstrackr accepts an abbreviated CSV file (title, abstract) and returns the same data back with a score field added. It will be the job of the I3 process to link this partial data back to the original record
  • The deduping library would output two files - a deduplicated library and a second (optional) output of the found duplicates

Example CLI usage:

The following is an example of the above process using the CLI. Since this tool is still in development the below is conceptual only but provides some insight into how the I3 implementation would work:

We assume that a workspace already exists with input-library.xml.

# Accept input-library.xml, dedupe and output to deduped.xml + dupes.xml
i3 --input input-library.xml --output deduped.xml --output dupes.xml --action iebh/dedupe

# Input deduped.xml, spider references by one generation forward/back output as spidered.csv
i3 --input deduped.xml --output spidered.csv --action iebh/spider --setting forward=1 -setting backward=1

# Input spidered.csv, launch and wait for Abstrackr functionality to compete, output as screened.xml EndNote library
i3 --input spidered.csv --output screened.xml --action brown/abstrackr

Notes:

  • The long form options (e.g. --input) are used in the above for readability, short form options (--input or -i) would be available also
  • The named action (e.g. --action iebh/dedupe) corresponds to a registered Docker image which would contain a manifest indicating how the container should be built and input/output processed
  • As the deduplication step provides two outputs the --output argument specifies two output filenames

Integration with a UI

Once the scenario testing stage has been finalized, production can begin on a more accessible user interface (UI). This would take the underlying API developed for the I3 CLI tool and provide a web page which would bring the tool components together.

Wider implementation of existing tools

In order to get wider circulation of the I3 platform more modules will need to be incorporated into its design to encourage other solution creators.

Back to contents