Figure out input dataset for the tool

UP2040499 / auto-osint-v

An automated tool for Validating OSINT. This forms part of the final step of OSINT production as detailed by NATO's open source handbook (2001). This is a research artefact for my Dissertation at the University of Portsmouth.

https://up2040499.github.io/auto-osint-v/

Creative Commons Zero v1.0 Universal

2 stars 0 forks source link

Figure out input dataset for the tool #1

Closed UP2040499 closed 1 year ago

UP2040499 commented 1 year ago

So the tool's input dataset will consist of a standardised set of "intelligence" that it will validate.

Will this intelligence have a source attached to it?
Should the tool try to create the intelligence (i.e. completing the 'processing' step of OSINT)?
Or are we creating the intelligence for it to validate - which could be just as time-consuming?

Perhaps the easiest method is to create intelligence statements for the tool to verify

Only issue could be the 'echo effect' where the intelligence may have been created from the same source supposedly corroborating it.

UP2040499 commented 1 year ago

Should statements have sources attached to them?

Intelligence statements could be single-sourced, multiple-sourced, or no sources at all (could be analogous to having formed the statement from classified intelligence so no public source would be provided).
This can be written into a CSV file where you have one column for the statement and one for the source(s) used in creating the statement. Other columns could also be added later on for testing or training such as labels for ambiguous, false, true or partly true statements.

UP2040499 commented 1 year ago

Example of what the csv could look like:

Statement	Source(s)
Statement One	link to Twitter thread, link to article
Statement Two	link to article
Statement Three

If multiple values in one field are comma-separated, like the sources for statement one above, use double quotes (") around the values (more).