not hypernormally connected

[arademaker]

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]

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.