Proposed Taxonomy of Flaw-related terms

[samm82]

After thinking about how we've I've been discussing syntactic flaws vs. semantic flaws, I've come up with a better (?) way of understanding these terms and therefore structuring this section of my thesis, where "Flaws and Related Terms" would be the new title of Sec. 2.1, with the following being subsections:

• Flaw: the general, "catch-all" term as we've been using it (after moving away from discrepancy as in Introduce "Flaw" as a catch-all for what we've called "Discrepancies" #140). Each flaw has a "manifestation" and a "domain", and flaws between multiple sources may be referred to as "inconsistencies".
  1. Manifestation: the specific way a flaw manifests; how information is wrong (previously called a "syntactic flaw").
  2. Domain (too overloaded): what information is wrong (previously called a "semantic flaw")
  3. Inconsistency: a synonym for "flaw" when discussing a "flaw between two sources", since "an inconsistency between two sources" flows more naturally. Additionally, "this clearly indicates that there is disagreement between the sources, but also does not imply that either one is correct—the inconsistency could be with some ground truth if neither source is correct!" (my thesis, Sec. 2.1). We could also make the case that "flaw" and "inconsistency " are synonyms, as the "inconsistency " could be with some ground truth, but make clear that we prefer the term "flaw" since it is "stronger" as we've previously discussed.

Regardless of the specifics, I think this is a good structure for introducing this terminology, so I'll continue to progress on #140 while keeping these in mind but not baking them in until we come to a consensus.

[smiths]

@samm82 I like where you are going with this and the systematic way you are considering your options. The term flaw as the catch-all and the term discrepancy as a synonym in the relevant situations makes sense to me. I'm less sure about manifestation and domain. Can you give some examples of each? Having domain be about "what" knowledge area has a flaw sounds good. Defining manifestation using both "how" and "why" seems odd.

[samm82]

Current manifestations: mistakes, omissions, contradictions, ambiguities, overlaps, and redundancies.
Current domains: approach categories, synonym relations, parent-child relations, definitions, terms themselves (i.e., "names" or "labels"), and citations (or "traceability").

I think I've been using "how" and "why" to ask the same question, when really, they ask slightly different ones. I'm really looking at how a flaw manifests; the why would more be at the level of "why was the mistake made", "who was responsible", etc. Just another example of how terminology can be imprecise (since "how" and "why" can both mean "for what reason"; in Japanese, both なぜ and どうして can be translated to either of "how" or "why" with different formality levels).

[smiths]

@samm82 I can see the confusion around "how" and "why". I didn't know that fact about the Japanese language. Neat! The distinction that manifestation is "how there is a flaw" and domain is "what information is flawed" sounds good to me. I'm still not sure of the manifestation and domain as the terms we use, but the distinction between the concepts seems good to me. 😄 How the flaw is manifested seems pretty good as a dimension of describing a flaw. I think I like manifestation. Domain is more problematic because the domain is your domain of testing terms, but domain is used for so many other purposes. Could we talk about the "type" of the flaw, instead of the "domain" of the flaw?

[samm82]

Domain is more problematic because the domain is your domain of testing terms, but domain is used for so many other purposes.

I'm not sure what you mean by the first part of this (do you mean "domain" as in "scope"?), but definitely agree with the second! I think this is a rare example where reusing an existing term would provide clarity instead of confusion, since "synonym relations" comprise one "domain" of our analysis. We could also use the word "area" which is more vague but less overloaded (SWEBOK uses the concept of "knowledge areas" slightly differently). Looking through synonyms doesn't really reveal anything I think is a good fit…

Could we talk about the "type" of the flaw, instead of the "domain" of the flaw?

We could, but we already define "test type" as an approach categorization which would lead to overloading this term. I try to (even during our interactions!) stick to using the word "kind" when I mean it generically, reserving "type" to refer to the category of test approaches.

[smiths]

We discussed this during our recent meeting (#156). We thought maybe we could talk about a flaw's location, rather than its domain. I agree that type is a horrible option. 😄