Open goodmami opened 9 years ago
A phrase like "顔が明るくなる" yields the following MRS:
[ TOP: h0 INDEX: e2 [ e TENSE: past MOOD: indicative PROG: - PERF: - ASPECT: default_aspect PASS: - SF: prop ] RELS: < [ udef_q_rel<0:1> LBL: h4 ARG0: x3 [ x PERS: 3 ] RSTR: h5 BODY: h6 ] [ "_kao_n_rel"<0:1> LBL: h7 ARG0: x3 ] [ "_akarui_a_1_rel"<4:7> LBL: h8 ARG0: e2 ARG1: x3 ] [ "_naru_v_rel"<8:10> LBL: h1 ARG0: e2 ARG1: h8 ] > HCONS: < h0 qeq h1 h5 qeq h7 > ]
Note that _akarui_a_1_rel and _naru_v_rel both have the same ARG0, which is invalid (since one isn't a quantifier of the other). In this other parse, _kao_n_rel and its quantifier are detached from the rest of the MRS:
_akarui_a_1_rel
_naru_v_rel
_kao_n_rel
[ TOP: h0 INDEX: e2 [ e TENSE: past MOOD: indicative PROG: - PERF: - ASPECT: default_aspect PASS: - SF: prop ] RELS: < [ udef_q_rel<0:1> LBL: h4 ARG0: x3 [ x PERS: 3 ] RSTR: h5 BODY: h6 ] [ "_kao_n_rel"<0:1> LBL: h7 ARG0: x3 ] [ "_akarui_a_1_rel"<4:7> LBL: h8 ARG0: e9 [ e TENSE: untensed MOOD: indicative PROG: - PERF: - ASPECT: default_aspect PASS: - SF: prop ] ARG1: i10 ] [ "_naru_v_3_rel"<8:10> LBL: h1 ARG0: e2 ARG1: h11 ] > HCONS: < h0 qeq h1 h5 qeq h7 h11 qeq h8 > ]
I have fixed the issue with things having the same ARG0, but the semantics is still not quite right. naru-hiragana stem needs some work.
A phrase like "顔が明るくなる" yields the following MRS:
Note that
_akarui_a_1_rel
and_naru_v_rel
both have the same ARG0, which is invalid (since one isn't a quantifier of the other). In this other parse,_kao_n_rel
and its quantifier are detached from the rest of the MRS: