search

CognitiveAtlas / ontology

Repository for the ontology and discussion/issue with ontology content
MIT License
4 stars 1 forks source link

Possible assertion type "is equivalent to" (or maybe "does not exist")? #10

Open tsalo opened 6 years ago

tsalo commented 6 years ago

I was thinking that it might be useful if contributors could assert that two terms are the same, or that one term doesn't exist.

I have an example, although I'm not exactly an expert on memory. Currently in the Cognitive Atlas, explicit memory is considered separate from, and a kind of, declarative memory. Most of the literature I've read conflates the two, so, assuming the differentiation in the Cognitive Atlas isn't an error, I think this reflects two competing theories. In order to compare the two theories, we need to be able to assert that the two terms are the same, as described by "theory A", since having them separated and connected with the kind-of relationship is part of "theory B".

poldrack commented 6 years ago

thanks - I guess I had envisioned using the synonym function to express that two concepts were the same. but the idea of explicitly expressing that a concept does not exist is interesting. it seems that this would need to be a relation between a theory and a concept, right? e.g. "asserts_nonexistence(theoryA,conceptX)". does that sound right?

On Thu, Jan 4, 2018 at 10:58 AM, Taylor Salo notifications@github.com wrote:

I was thinking that it might be useful if contributors could assert that two terms are the same, or that one term doesn't exist.

I have an example, although I'm not exactly an expert on memory. Currently in the Cognitive Atlas, explicit memory https://cognitiveatlas.org/concept/id/trm_4a3fd79d0a281/ is considered separate from, and a kind of, declarative memory https://cognitiveatlas.org/concept/id/trm_4a3fd79d0a04f/. Most of the literature I've read conflates the two, so, assuming the differentiation in the Cognitive Atlas isn't an error, I think this reflects two competing theories. In order to compare the two theories, we need to be able to assert that the two terms are the same, as described by "theory A", since having them separated and connected with the kind-of relationship is part of "theory B".

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/CognitiveAtlas/ontology/issues/10, or mute the thread https://github.com/notifications/unsubscribe-auth/AA1KkIG8yIGq8iWicMBEcZ-TvUdaIQu3ks5tHR9dgaJpZM4RTa-z .

-- Russell A. Poldrack Albert Ray Lang Professor of Psychology Professor (by courtesy) of Computer Science Bldg. 420, Jordan Hall Stanford University Stanford, CA 94305

poldrack@stanford.edu http://www.poldracklab.org/

tsalo commented 6 years ago

If synonym assertions can be tied to a theory or reference, that would work perfectly.

At least for me (and I might just not be doing things the best way), when I do text mining with the Cognitive Atlas, I extract each term and its synonyms and count them all toward the term. When there are duplicates (e.g., the synonym of one term is the same as the preferred name of another term), I use the term that is closest to the duplicate. So in the case of explicit memory and declarative memory, whichever theory it is that posits that the two concepts are distinct will win out, even if explicit memory is included as a synonym of declarative memory. If the synonym's existence is tied to some number of papers, though, I could just choose which term to count the string towards based on which one has more citations, or something.

Of course, this is only important for text extraction and probably wouldn't affect people who are using the ontology as a reference; although, without some way of clarifying that the synonym assertion stands in contrast to the assertion that explicit memory is a kind of declarative memory, I can see those users being confused as well.

Tying the nonexistence assertion between a theory and a concept makes a lot of sense of me. I hadn't considered including theories directly in assertions, but it's a cool idea.

Alternatively, would it be possible to expose assertions and the terms themselves to users for voting? I think it would be useful to be able to characterize disagreement in the field, even if votes require references. E.g., instead of asserting that a conceptX doesn't exist (based on some paper), I could go to the concept and add in a reference for the paper against the concept.

poldrack commented 6 years ago

yes, great idea re: voting, though implementation of that may take a little while.

On Thu, Jan 4, 2018 at 11:30 AM, Taylor Salo notifications@github.com wrote:

If synonym assertions can be tied to a theory or reference, that would work perfectly.

At least for me (and I might just not be doing things the best way), when I do text mining with the Cognitive Atlas, I extract each term and its synonyms and count them all toward the term. When there are duplicates (e.g., the synonym of one term is the same as the preferred name of another term), I use the term that is closest to the duplicate. So in the case of explicit memory and declarative memory, whichever theory it is that posits that the two concepts are distinct will win out, even if explicit memory is included as a synonym of declarative memory. If the synonym's existence is tied to some number of papers, though, I could just choose which term to count the string towards based on which one has more citations, or something.

Of course, this is only important for text extraction and probably wouldn't affect people who are using the ontology as a reference; although, without some way of clarifying that the synonym assertion stands in contrast to the assertion that explicit memory is a kind of declarative memory, I can see those users being confused as well.

Tying the nonexistence assertion between a theory and a concept makes a lot of sense of me. I hadn't considered including theories directly in assertions, but it's a cool idea.

Alternatively, would it be possible to expose assertions and the terms themselves to users for voting? I think it would be useful to be able to characterize disagreement in the field, even if votes require references. E.g., instead of asserting that a conceptX doesn't exist (based on some paper), I could go to the concept and add in a reference for the paper against the concept.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/CognitiveAtlas/ontology/issues/10#issuecomment-355375848, or mute the thread https://github.com/notifications/unsubscribe-auth/AA1KkMVAvfLJRFvb8DOiA_k44EN88JZeks5tHSbFgaJpZM4RTa-z .

-- Russell A. Poldrack Albert Ray Lang Professor of Psychology Professor (by courtesy) of Computer Science Bldg. 420, Jordan Hall Stanford University Stanford, CA 94305

poldrack@stanford.edu http://www.poldracklab.org/