Skip to content

Comment on CauseNet: Towards a causality graph extracted from the webparent

Comments

Koller and Friedman write in "Probabilistic Graphical Models" about the "clarity test", so that state variables should be clear for an all seeing observer.

States like "human_activity" are not objectively measurable.

Fairly PGMs and causal models are not the same, but this way of thinking about state variables is an incredible good filter.

States like "human_activity" are not objectively measurable.

Well, or at least they would need a heavy dose of operationalisation.

AboutSource Built by g1lg1l

Hackerly is an independent reader for Hacker News, built on the public HN API. Not affiliated with Y Combinator.