[ LTOP: h0 INDEX: e2 [ e SF: prop TENSE: past MOOD: indicative PROG: - PERF: - ] RELS: < [ _some_q_indiv0:4 LBL: h4 ARG0: x3 [ x PERS: 3 NUM: sg IND: + ] RSTR: h5 BODY: h6 ] [ _big_a_15:8 LBL: h7 ARG0: e8 [ e SF: prop TENSE: untensed MOOD: indicative PROG: bool PERF: - ] ARG1: x3 ] [ _dog_n_19:12 LBL: h7 ARG0: x3 ] [ _chase_v_113:19 LBL: h1 ARG0: e2 ARG1: x3 ARG2: x9 [ x PERS: 3 NUM: sg IND: + ] ] [ _every_q20:25 LBL: h10 ARG0: x9 RSTR: h11 BODY: h12 ] [ _cat_n_126:29 LBL: h13 ARG0: x9 ] > HCONS: < h0 qeq h1 h5 qeq h7 h11 qeq h13 > ICONS: < > ]
h1:+every [RSTR h2] h3:+cat {h2 qeq h3}
SENT: kitty probably sleeps [ LTOP: h0 INDEX: e2 [ e SF: prop TENSE: pres MOOD: indicative PROG: - PERF: - ] RELS: < [ proper_q0:5 LBL: h4 ARG0: x3 [ x PERS: 3 NUM: sg IND: + ] RSTR: h5 BODY: h6 ] [ named0:5 LBL: h7 CARG: “Kitty” ARG0: x3 ] [ _probable_a_16:14 LBL: h1 ARG0: i9 ARG1: h10 ] [ _sleep_v_115:21 LBL: h11 ARG0: e2 ARG1: x3 ] > HCONS: < h0 qeq h1 h5 qeq h7 h10 qeq h11 > ICONS: < > ]
h1:+probable[ARG1 x] y:/v {y qeq x}
(p*[A* y] | h:try_v*[ARG1 x]) n:v*[A* c] /v*[ARG* x] quarterly[ARG1 x] x:+result x:=1 y:crisis[] h1:+terrible[ARG* x1]
Exemplo:
h1:+terrible[ARG1 y]
(h2:+terrible[ARG1 x] | h3:run )
{ h1 qeq h3 }
“12 + 34*56”?
=> “12” “+” “34” “*” “56”
Grammatica EBNF
EXP := EXP OP EXP | NUM OP := “+” | “*” NUM := [0-9]+
=> “1” “2” “+” …
Gramatica BNF
EXP := EXP OP EXP | NUM OP := “+” | “*” NUM := D NUM | _ D := 0 | 1 | … | 9
“a b”? => “a” “b” => “a” ” ” “b”
So how to tokenize?
h:quarterly[ARG1 x]
h:quarterly[ ARG1 x ] // 1 predicado h:quarterly [ ARG1 x ] // 2 predicados
Following W3C’s EBNF notation
Wql := Space? PredExp Space? ("{" Space? HList? Space? "}")? Space?
PredExpr := Predication
| PredExpr Space? "|" Space? PredExpr
| PredExpr Space PredExpr
| "(" Space? PredExpr Space? ")"
| "!" Space? PredExpr
Predication := "^"? (Var ":")? PredMod? PredPat ("[" Space? ArgList? Space? "]")?
| (Var ":")? "[" Space? ArgList? Space? "]"
ArgList := Argument | Argument Space? "," Space? ArgList
Argument := RolePat (Space Var)?
HList := Heq | Heq Space? "," Space? Hlist
Heq := Var Space? HconsOp Space? Var
HconsOp := "=q"
PredPat := "_"? [^\s_]+ (_[nvajrscpqxud])? (_([^\s_<]|<(?![-0-9:#@ ]*>\s))+)? (_rel)?
RolePat := [a-zA-Z0-9\*]+
Var := [a-zA-Z] [a-zA-Z0-9]*
PredMod := [+=/]
Space := [\s]+
run[ARG1 x]
:e1 mrs:predicate “run” . :e2 mrs:arg1 :e3 :e2 mrs:role :e3
[* x] =>
– mais literal select ?x0 ?x0 a mrs:MRS . ?x0 mrs:hasEP ?x1 . ?x1 ?p ?x2 . ?p rdfs:label ?l filter regex(?l, “.*”) }
– mais pre-processamento select ?x0 ?x0 a mrs:MRS . ?x0 mrs:hasEP ?x1 . ?x1 mrs:role ?x2 . }
Semântica das predicações:
And (And P (Not Q)) (Or R S) [Hcons h m]
operacoes:
h1:every*[RSTR h2] h3:+cat {h2 qeq h3}
from ?g { { ?x00 mrs:predicate ?p ?x00 mrs:hole ?h1 ?x00 mrs:rstr ?h2 filter regex(?p, “every.*”) }
{ ?x02 mrs:predicate/mrs:lemma “cat” ?x02 mrs:hole ?h3 }
}
{ ?m a mrs:MRS ?m mrs:hasEP ?x00 ?m mrs:hasEP ?x02 ?m mrs:hasHcons ?mhs }
{ ?x00 mrs:predicate ?p ?x00 mrs:hole ?h1 ?x00 mrs:rstr ?h2 filter regex(?p, “every.*”) }
{ ?x02 mrs:predicate/mrs:lemma “cat” ?x02 mrs:hole ?h3 }
{ ?mhs mrs:left ?h2 ?mhs mrs:right ?h3 ?mhs a mrs:QEQ }
f :: Wql -> String ? f :: Wql -> Sparql q :: Sparql -> String
f :: PredExp -> Sparql f (p | q) = Graph (f p) UNION Graph (f q)
Not (Not P) == P Not (And P Q) == Or (Not P) (Not Q) Not (Or P Q) == And (Not P) (Not Q)
pros? cons?
- http://sdp.delph-in.net/2015/search.html
- http://wsi.mybluemix.net/demo/
- http://moin.delph-in.net/WeSearch/QueryLanguage
- http://lists.delph-in.net/archives/developers/2020/003199.html
- http://moin.delph-in.net/wiki/PredicateRfc
- https://hackage.haskell.org/package/base-4.14.1.0/docs/Text-ParserCombinators-ReadP.html
- http://hackage.haskell.org/package/parsec
- http://www.macs.hw.ac.uk/~rs46/slides/rdf4h/slides.html
- https://github.com/robstewart57/hsparql
- https://hackage.haskell.org/package/swish