Clues Business Domain

[ryjen]

Clues need to have point values associated with them for the game and from a business perspective.

Some clues maybe superflous in nature only as a guide the user UX (such as color or rhyme)

While other clues maybe be core to the underlying functionality (such as matching image embeddings)

Clues may have logic determining their individual point value as higher or lower than the domain level standard (such as the machine vision confidence rating of labels).

The domain must be able to choose the best clue from a group of clues, called proofs.

Tasks:

• [ ] overall color (nearness to color name)
• [ ] segmented or object color
• [ ] overall label
• [ ] segmented or object label
• [ ] user defined label (voice or text)
• [ ] specific location (accuracy, terrain)
• [ ] #48
• [ ] clue point logic (number of clues)
• [ ] best clue logic / sorting

Optional (another sprint)

• [ ] location in geofence

[ryjen]

Need to update data model for additional fields for phase 1 sub-analyzers:

segmentations and objects:

• labels
• colors

[ryjen]

Some clues seem to be better as post-selection analysis.

For example:

• embeddings/matching on final image
• rhyming for selected labels