Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

not hypernormally connected #41

Open
arademaker opened this issue Mar 31, 2023 · 1 comment
Open

not hypernormally connected #41

arademaker opened this issue Mar 31, 2023 · 1 comment

Comments

@arademaker
Copy link
Member

arademaker commented Mar 31, 2023

Consider the sentence from the SICK entailment dataset:

This group of people is practicing water safety and wearing preservers

One of the readings produce the MRS:

[ TOP: h0
  INDEX: e2 [ e SF: prop TENSE: pres MOOD: indicative PROG: + PERF: - ]
  RELS: < [ _this_q_dem<0:4> LBL: h4 ARG0: x3 [ x PERS: 3 NUM: sg IND: + ] RSTR: h5 BODY: h6 ]
          [ _group_n_of<5:10> LBL: h7 ARG0: x3 ]
          [ _of_p<11:13> LBL: h7 ARG0: e8 [ e SF: prop TENSE: untensed MOOD: indicative PROG: - PERF: - ] ARG1: x3 ARG2: x9 [ x PERS: 3 NUM: pl IND: + ] ]
          [ udef_q<14:20> LBL: h10 ARG0: x9 RSTR: h11 BODY: h12 ]
          [ _people_n_of<14:20> LBL: h13 ARG0: x9 ARG1: i14 ]
          [ _practice_v_1<24:34> LBL: h1 ARG0: e2 ARG1: x3 ARG2: x15 [ x PERS: 3 NUM: pl IND: + ] ]
          [ udef_q<35:70> LBL: h16 ARG0: x15 RSTR: h17 BODY: h18 ]
          [ compound<35:59> LBL: h19 ARG0: e20 [ e SF: prop TENSE: untensed MOOD: indicative PROG: - PERF: - ] ARG1: x15 ARG2: x21 [ x PERS: 3 NUM: sg IND: - ] ]
          [ udef_q<35:59> LBL: h22 ARG0: x21 RSTR: h23 BODY: h24 ]
          [ compound<35:47> LBL: h19 ARG0: e25 [ e SF: prop TENSE: untensed MOOD: indicative PROG: - PERF: - ] ARG1: x21 ARG2: x26 ]
          [ udef_q<35:40> LBL: h27 ARG0: x26 RSTR: h28 BODY: h29 ]
          [ _water_n_1<35:40> LBL: h30 ARG0: x26 ]
          [ _safety_n_1<41:47> LBL: h19 ARG0: x21 ]
          [ _and_c<48:51> LBL: h31 ARG0: e32 [ e SF: prop ] ARG1: e20 ARG2: e33 [ e SF: prop TENSE: untensed MOOD: indicative PROG: + PERF: - ] ]
          [ _wear_v_1<52:59> LBL: h31 ARG0: e33 ARG1: x15 ARG2: i34 ]
          [ _preservers/nns_u_unknown<60:70> LBL: h31 ARG0: x15 ] >
  HCONS: < h0 qeq h1 h5 qeq h7 h11 qeq h13 h17 qeq h31 h23 qeq h19 h28 qeq h30 > ]

This MRS passed in the Pydelphin well-formed test:

https://pydelphin.readthedocs.io/en/latest/api/delphin.mrs.html#delphin.mrs.is_well_formed

But Utool reports:

A semantic error occurred while decoding the graph. The dominance children h1 and h19 of the root h16 are not hypernormally connected with each other.

Is that a possible situation? The question is if all well-formed MRS should be solvable by Utool or not. Or is it a problem in the Pydelphin well-formed test?

@goodmami
Copy link
Member

Utool has a slightly different notion of what is well-formed. PyDelphin's check is not terribly strict; the documentation you linked to even states that it checks 3 things:

  • is the MRS connected (in a graphical sense)
  • does it have the intrinsic variable property
  • does it plausibly scope (a handful of fairly simple checks)

The meanings of these things are further explained in the documentation. If you can think of any further well-formedness tests, then I'd consider adding them to PyDelphin.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants