Why are these two MRSs not isomorphic?

MRS
1

I have one MRS I got from parsing the phrase “cute cat” with the ERG and I have another that I constructed from scratch, they look isomorphic to me, but PyDelphin doesn’t recognize them as such. What am I missing?

ERG MRS
[ 
    LTOP: h0 
    INDEX: e2 [ e SF: prop ] 
    RELS: < 
        [ unknown<0:8> 
            LBL: h1 
            ARG0: e2 
            ARG: x4 [ x PERS: 3 NUM: sg IND: + ] 
        ]  
        [ udef_q<0:8> 
            LBL: h5 
            ARG0: x4 
            RSTR: h6 
            BODY: h7 
        ]  
        [ _cute_a_1<0:4> 
            LBL: h8 
            ARG0: e9 [ e SF: prop TENSE: untensed MOOD: indicative PROG: bool PERF: - ] 
            ARG1: x4 
        ]  
        [ _cat_n_1<5:8> 
            LBL: h8 
            ARG0: x4 
        ] > 
    HCONS: < h0 qeq h1 h6 qeq h8 > 
    ICONS: < > 
]

My MRS:
[ 
    TOP: h12 
    INDEX: e9 
    RELS: < 
        [ unknown 
            LBL: h11 
            ARG: x3 
            ARG0: e9 
        ]
        [ udef_q 
            LBL: h8 
            ARG0: x3 
            RSTR: h6 
            BODY: h7 
        ] 
        [ _cute_a_1 
            LBL: h2 
            ARG0: e0 
            ARG1: x3 ] 
        [ _cat_n_1 
            LBL: h2 
            ARG0: x3 
        ] > 
    HCONS: < h6 qeq h2 h12 qeq h11 > 
]

The key things I see in the ERG MRS are…

  • TOP =q unknown_LBL
  • INDEX = unknown_ARG0
  • RSTR of udef_q =q cute_LBL/cat_LBL
  • unknown_ARG = udef_q_ARG0 = cute_ARG1 = cat_ARG0

All of these conditions are met in my MRS, but it still isn’t working. Is anything jumping out?

EDIT: It seems that the fact that the ERG version includes additional information about some variables (tense, person, number, etc) is what’s preventing it from being isomorphic. … still, the ERG won’t generate from my MRS, which seems odd because other than the lack of that information it’s still “correct” (as far as I can tell)

2

Hi,

your MRS does not seem to have any of the properties, such as [ e SF: prop ] on the INDEX.

These convey part of the meaning, so need to be generated as well.

I think your MRS probably subsumes the ERG’s MRS, …

1 Like
3

Yes, thanks that was causing them to not be isomorphic!
There was another unrelated error preventing generation, but I got it to generate after all.
It does generate “cute cats” as well since I didn’t specify that it’s singular, but I’m okay with that and adding an option in my program to constrain if the user wants.

4

Sounds like you got it. You can disable checking of morphosemantic properties with properties=False in the is_isomorphic() call. Also the documentation says this:

Isomorphicity compares the predicates of a semantic structure, the morphosemantic properties of their predications (if properties=True), constant arguments, and the argument structure between predications. Non-semantic properties like identifiers and surface alignments are ignored.

And if I recall correctly, in order to generate you had to have some minimum set of properties specified. I think it’s only TENSE on eventualities.

1 Like