mhtess / comparison-class Goto Github PK

Rather than the current implementation, which mixes at every node, thus discounting the leaves with many intermediate nodes.

This may or may not be useful

e.g., a long movie is like 2 hours 30 minutes long

This will involve:

• A scalar adjective (e.g. tall, expensive, ...) [see linguistics papers for ideas]
• An entity (e.g. Bob)
• Information about the entity (bob is a giraffe, a 4 y.o.)

This probably involves adding entities to the hierarchy

First case: "John is tall" (knowing John is a person)

• grab prior on height for humans
• incorporate with adjectives model, though I think we will want to write it more in the style of the generics model

At the beginning of the listener (and prior) experiments, tell participant they will be asked to estimate quantities of time, distance, weight, temperature and money. Then ask them their preferred units for each, make slides with those units (no more boxes). For money, tell them they must responds in US dollars.