Human error in triples validation and solution

Hello all,
I would like to review a case that touched me in the preparation of triples and offer a possible solution to this problem.

Problem: Human error in triples validation
When checking triples encountered different interpretations from validators about whether or not to count triples. This is not a case where consensus must be reached within the same case, and I’m talking about what is generally similarly made triples, but with different data.

Example:
Triple #: Person “A” + his profession + reference to resource “A” confirmation. Result: +5 points
Triple #2: Person “B” + his profession + link to resource “A” confirmation. Result: -5 points

Solution: Make the “rules of the game” transparent
In my opinion, the solution may be the following, to make the rules a certain minimum of information about the person, company, etc., which should be set out in the resource that he could be considered admissible to confirm the identity or existence of the company, product, etc. If the criteria are met, then the checker should count the triple, if not, then not.

In my opinion, the proposed approach will reduce tension and uncertainty in this matter.

Thank you all :slightly_smiling_face:

8 Likes

Sounds reasonable for me!

A strange suggestion that doesn’t work for several reasons. As I understand it, you want the resource that you are validating to have a certain minimum of information in it?

But what if the person being verified does not cope well with filling his/her media/other resources with information? It will not be recognized only because of these nuances?

And the biggest problem now is the numerous attempts to bypass the validation system. Some time ago, solid failures were validated, for the past few days only acceptances have been validated, despite the correctness and incorrectness of the data.

The only thing that can help in validation is adding the votes of not only tester validators, but also team members, as well as artificial intelligence, which is impossible in the current realities.

Hey! If I have understood correctly, your general idea is that a certain minimum of information should be required in order for a triple to be placed in the validation queue (not just name and predicate)? If it is the case, then I fully support it and pointed that out several times in Discord.
PS Sorry if I did not get u right

hey!

The key idea is to define the rules by which a participant can understand in which case his triple will be counted and in which case it will not. To do this, he must understand what information, for example, about the person about whom he wants to write should be contained in this or that resource to which he refers when confirming the “reality” of this person.

For example: date of birth, photo, company in which the person works, biography, main works, etc.

Now it just turns out that if the site that confirms the identity is not facebook, twitter, linkedln etc. there is a great risk that triple will be rejected

What you are talking about a little different, but it works on the same side - should be defined responsibility not only checker, but also the one who uploads the data.

1 Like

Not really, I propose that the person who uploads triple clearly understood that to confirm the identity of some (for example) person, the site to which he refers to contains a minimum of information about this person: date of birth, city, photo, achievements, etc. (the list is negotiable).

If this data is not available, then there is no need to upload such a triple and waste your time and the time of the checker.