Fun retrospective discussion links:
Trust and Authority - they are not the same thing [webmasterworld.com], September 2008.
How To Establish EAT (Expertise, Authority, Reputation) [webmasterworld.com], August 2014.
Bing: Trust and Authority Usually Sees Higher Rankings [webmasterworld.com], October 2014.
Google's Knowledge Graph:Knowledge-Based Trust: Estimating the Trustworthiness of Web Sources [webmasterworld.com], March 2015.
How Google can determine trust [webmasterworld.com], May 2016.
Google Adds Trust Project Labelling [webmasterworld.com], November 2017.
E-A-T-ing well now even more critical [webmasterworld.com], May 2019.
While my broader thoughts on the matter of how EAT might be implicitly determined by an algorithm are mentioned in some of the above, given the explicit question and the new day with my feet up looking for an excuse not to mow the grass...
Note: an important disclaimer is that each of the following has inherent weakness, is game-able; the critical flaw in implicit approximation of explicit goals.
Expertise: expert skill or knowledge in a particular field.
* credentials aka awards/certification/licensing in field.
* career aka number of years working in field.
* performance aka record of publication/speaking on field.
* peer respect aka number of citations, mentions by others in field.
Authority: power to influence others because of recognised expertise/knowledge in field.
* expertise, see above
* legal/rational institutions in field, i.e. professional associations, research universities, government oversight/regulating organisations.
* link graph nodes of exceptional authority aggregation in field/niche/vertical.
Trust: firm belief in the reliability, truth, ability, or strength of an entity.
* credibility in field aka authority (see above).
* reliability aka testimonials, mentions.
* safety/security aka level of confidentiality (eg PII), ethical/legal/social accolades versus complaints.
* focus aka is interest in field total or tangential, professional or hobby, etc.
The foregoing is largely field/niche/vertical oriented; I suspect there is a probable broader website accessibility/usability perspective input as well although such always seem to be minor ‘tiebreak’ level influences (probably because the greater their weight the greater the impact on ‘too big to exclude’ sites).
Dangnabit the grass didn’t mow itself.