Add documentation from the semantic relations

[fcbond]

It would be good to use the names, short descriptions and possibly link to the semrel documentation.

1. We need to have the semrel documentation in github
2. It should be visible somewhere (in GWA or OMW (or both)) so we can link to it

[arademaker]

Added @viguardieiro

[fcbond]

Currently the sense-level symbol dictionary is inlined in the omw_concept.html template:

 {% set sl2sym = {1:'⇔', 4:'⊞', 5:'⊳'}  %}

We only have symbols for antonym, pertainym, and derivation. This should be replaced and extended by the docs.

[goodmami]

Documentation is live: https://globalwordnet.github.io/gwadoc/

But it still needs work to be filled in, and currently there is no way to select a different language (although we need more data in other languages for that to be relevant).

[goodmami]

@fcbond pertainym and derivation are not described in gwadoc. Neither is participle (which is mentioned in OMW's srel.tab). I thought I had heard that pertainym was superseded by another relation, like also or other or something, but now I don't see any record of that, so perhaps I imagined it. So what is the current status of these 3 relations?

[fcbond]

gwadoc currently only has the synset-synset relations. Although there is some overlap in names, I think it is better for the sense-sense and sense-synset relations to be clearly different. ``` xml <!ATTLIST SenseRelation target IDREF #REQUIRED relType (antonym|also|participle|pertainym|derivation|domain_topic|has_domain_topic|domain_region|has_domain_region|exemplifies|is_exemplified_by|similar|other) #REQUIRED ``` I am not sure the best way of adding them to gwadoc. All three should be documented (but are not yet) I just checked and we are missing a few things from the database: 324 also_sees 15 region_domains 10 topic_domains 398 usage_domains 1055 syntactic_marker 2 verb_groups I will start another issue (#63) NLTK does not read participles, so we are also missing them *_domain links to lemmas, but I think it should link to synsets syntactic markers are for adjectives: (p) predicate position (a) prenominal (attributive) position (ip) immediately postnominal position We have no examples of participles (but e.g. elapsed/elapse should be), similar or other.

…

On Tue, Mar 26, 2019 at 4:36 PM Michael Wayne Goodman < ***@***.***> wrote: @fcbond <https://github.com/fcbond> pertainym and derivation are not described in gwadoc. Neither is participle (which is mentioned in OMW's srel.tab). I thought I had heard that pertainym was superseded by another relation, like also or other or something, but now I don't see any record of that, so perhaps I imagined it. So what is the current status of these 3 relations? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#33 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABD8xlIhCLfLydwvhq6SMpDpIkctI47Lks5vadwjgaJpZM4T8pwp> .

-- Francis Bond <http://www3.ntu.edu.sg/home/fcbond/> Division of Linguistics and Multilingual Studies Nanyang Technological University

[goodmami]

So to be clear, this is the list of sense-sense relations?

• antonym
• also
• also_sees
• participle
• pertainym
• derivation
• domain_topic
• has_domain_topic
• topic_domains
• domain_region
• has_domain_region
• region_domains
• exemplifies
• is_exemplified_by
• usage_domains
• similar
• other
• syntactic_marker
• verb_groups

And maybe also several more from #63 ?

What about sense-synset relations?

[fcbond]

So to be clear, this is the list of sense-sense relations? - antonym - also - participle - pertainym - derivation - domain_topic - has_domain_topic - domain_region - has_domain_region - exemplifies - is_exemplified_by - similar - other - syntactic_marker - verb_groups

also_sees is also topic_domains is domain_topic region_domains is domain_region and usage_domains is exemplifies And maybe also several more from #63

<#63> ? What about sense-synset relations?

domain_topic domain_region exemplifies and I guess 'other', just in case?

…

-- Francis Bond <http://www3.ntu.edu.sg/home/fcbond/> Division of Linguistics and Multilingual Studies Nanyang Technological University

[goodmami]

Ok

and I guess 'other', just in case?

How about we leave it out and add it if there's a need. It's not hard to add another relation to gwadoc; it's harder to recognize that one is unnecessary.

[fcbond]

Good idea.

…

On Wed, Apr 3, 2019 at 4:05 PM Michael Wayne Goodman < ***@***.***> wrote: Ok and I guess 'other', just in case? How about we leave it out and add it if there's a need. It's not hard to add another relation to gwadoc; it's harder to recognize that one is unnecessary. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#33 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABD8xoJoWnUSDxXse_dHVgrCr1BKWuadks5vdGDggaJpZM4T8pwp> .

-- Francis Bond <http://www3.ntu.edu.sg/home/fcbond/> Division of Linguistics and Multilingual Studies Nanyang Technological University

[goodmami]

I began adding these to gwadoc under a separate namespace (e.g., sense_sense_rels.antonym) but realized that the flat list we defined does not fit in the hierarchy we have for synset_synset_relations. E.g., we'd have to include constitutive, other, and domain for them to fit somewhere (and define new branches for the sense-only relations). But maybe if they are in a separate namespace they fit in a different hierarchy? But also how do we deal with reverse relations?

It makes me wonder, again, why we need to keep these in separate namespaces at all, instead of just putting some form.sense_sense=True, form.synset_synset=True flags to show their applicability. Do we expect project names to vary depending on the source or target type of a relation? For things like examples, can't we just annotate them to indicate which kind it is an example of (e.g., "this is a sense--synset example for relation x")?

[fcbond]

I think this is done for now.