Will be following this.
Honestly, similarity measures seems like a good start to build a more robust "bias" definition on, even if it won't be exactly the meaning I suggested above.
"Truth" is such a vague notion bordering on unusable in most rational contexts, as what most people consider "truth" contains some bias or other. When people quarrel over truthfulness is seldom the facts that they disagree on; it is the valuation of those facts in relation to the topic, i.e. relevance.
Not counting the type of bias which is not ashamed to twist the actual facts, but that is a rarity, and easily dismissed.
Eh, I think this now we are at a mostly semantic point: for example, there is nothing strange about arguing there exists some optimal weighting of various facts, and this optimal combination can then justifiably be labelled truth.
> Not counting the type of bias which is not ashamed to twist the actual facts, but that is a rarity, and easily dismissed.
Isolated instances are easily dismissed, but the USA is now ruled by a systematic dismissal of truth so it is not quite so easy.
Unbiased news and journalism should present the facts without a narrative, but this is very dry and doesn't get the same clicks as a breath taking story.
All I can say is that there are stories where the victims clothes matter greatly (hunting "accident") and others where it doesn't (wire fraud).