Skrypt do trenowanie modelu w oparciu o frame oraz slots, wstepny skrypt do ewaluacji evaluate.py, skrypt do testow

conllu_generator.py

@@ -0,0 +1,15 @@
+print("Script to automatically append data to data/dialog.conllu")
+print("Start typing now. Press Ctrl+C to stop.")
+
+while True:
+    with open("data/train_dialog.conllu", "a") as f:
+        text = input("Text: ")
+        act = input("Intent: ")
+        slots = text.split(" ")
+        f.write(
+            f"\n# text: {text}\n# intent: {act}\n# slots:\n"
+        )
+        for i, slot in enumerate(slots):
+            label = input(f"{i}/{slot} label: ")
+            f.write(f"{i+1}\t{slot}\t{act}\t{label}\n")
+        print("---")

data/dialog-04-01-01.tsv

@@ -6,7 +6,7 @@ user	a jakie są dostępne pizze?	request(menu)
 system	Mamy wiele rodzajów, czy ma Pan ochotę na coś konkretnego?	reqmore()
 system	plecam pizzę ananasową. sos pomidorowy, mozarella, ananas i cebula	offer(pizza=pineapple,ingredient=tomato_souce,ingredient=mozarella,ingredient=pineapple,ingredient=onion)
 user	świetnie, w takim razie poproszę dwie duże pizze diavola oraz margaritę	request(quantity=2,pizza=diavola)&request(pizza=margharita)
-system	oczywiście, czy chciałby Pan coś do picia?	affirm()&request(drink)							
+system	oczywiście, czy chciałby Pan coś do picia?	affirm()&request(drink)
 system	polecam sok z buraka	offer(drink=beetroot_juice)
 user	tak, poproszę jeden	affirm()&request(quantity=1)
 system	Oczywiście, płaci Pan kartą, blikiem, gotówką czy przelewem? Zamówienie na miejscu czy na wynos?	request(payment_method)&request(collection)

data/dialog-04-07-01.tsv

@@ -5,7 +5,7 @@ user	chciałbym zamówić pizzę wegetariańską w średnim rozmiarze	request(pi
 system	Oczywiście, czy chciałby Pan jakieś dodatki?	affirm()&reqmore(additional_ingredients)
 user	a co znajduje się domyślnie na takiej pizzy?	request(menu)
 system	ser, mozarella, sos pomidorowy, cebula oraz karczochy	offer(ingredients=cheese,ingredients=mozarella,ingredients=tomato_souce,ingredients=onion,ingredients=artichokes)
-user	dobrze, w takim razie odpowiada	affirm()									
+user	dobrze, w takim razie odpowiada	affirm()
 user	a rozmiar średni o ile jest mniejszy od dużego?	request(size_difference=medium-large)
 system	rozmiar średni: 30 cm	inform(medium_size=30)
 system	rozmiar duży: 40 cm	inform(large_size=40)

data/dialog-07-08-01.tsv

@@ -1,5 +1,5 @@
-kto	treść	treść
-user	Witam	welcomemsg()	
+kto	treść	akt
+user	Witam	welcomemsg()
 system	Dzień dobry, witam w naszej restauracji Pizzeria. W czym mogę pomóc?	welcomemsg()&request(order)
 user	Chcę jeść pickę	inform(food=pizza)
 system	Jaki rodzaj pizzy sobie Pan życzy, oraz jaki rozmiar?	request(type,size)

data/train_dialog.conllu

@@ -0,0 +1,772 @@
+# text: cześć
+# intent: welcomemsg 
+# slots: 
+1	cześć	welcomemsg	NoLabel
+
+# text: witam
+# intent: welcomemsg 
+# slots: 
+1	witam	welcomemsg	NoLabel
+
+# text: dzień dobry
+# intent: welcomemsg 
+# slots: 
+1	dzień	welcomemsg	NoLabel
+2	dobry	welcomemsg	NoLabel
+
+# text: dobry wieczór
+# intent: welcomemsg 
+# slots: 
+1	dobry	welcomemsg	NoLabel
+2	wieczór	welcomemsg	NoLabel
+
+# text: dziękuję
+# intent: bye
+# slots:
+1	dziękuję	bye	NoLabel
+
+# text: dzięki
+# intent: bye
+# slots:
+1	dzięki	bye	NoLabel
+
+# text: pozdrawiam
+# intent: bye
+# slots:
+1	pozdrawiam	bye	NoLabel
+
+# text: do widzenia
+# intent: bye
+# slots:
+1	do	bye	NoLabel
+2	widzenia	bye	NoLabel
+
+# text: dziękuję serdecznie
+# intent: bye
+# slots:
+1	dziękuję	bye	NoLabel
+2	serdecznie	bye	NoLabel
+
+# text: dziękuję bardzo
+# intent: bye
+# slots:
+1	dziękuję	bye	NoLabel
+2	bardzo	bye	NoLabel
+
+# text: papa
+# intent: bye
+# slots:
+1	papa	bye	NoLabel
+
+# text: tak
+# intent: affirm
+# slots:
+1	tak	affirm	NoLabel
+
+# text: poproszę
+# intent: affirm
+# slots:
+1	poproszę	affirm	NoLabel
+
+# text: to poproszę
+# intent: affirm
+# slots:
+1	to	affirm	NoLabel
+2	poproszę	affirm	NoLabel
+
+# text: okej
+# intent: affirm
+# slots:
+1	okej	affirm	NoLabel
+
+# text: okej to poproszę
+# intent: affirm
+# slots:
+1	okej	affirm	NoLabel
+2	to	affirm	NoLabel
+3	poproszę	affirm	NoLabel
+
+# text: poproszę pizzę hawajska w rozmiarze dużym
+# intent: inform/order
+# slots:
+1	poproszę	inform/order	NoLabel
+2	pizzę	inform/order	B-food
+3	hawajska	inform/order	B-pizza
+4	w	inform/order	NoLabel
+5	rozmiarze	inform/order	NoLabel
+6	dużym	inform/order	B-size
+
+# text: chciałbym zamówić pizzę
+# intent: request/menu
+# slots:
+1	chciałbym	request/menu	NoLabel
+2	zamówić	request/menu	NoLabel
+3	pizzę	request/menu	B-food
+
+# text: chciałbym pizzę z kurczakiem
+# intent: inform/order
+# slots:
+1	chciałbym	inform/order	NoLabel
+2	pizzę	inform/order	B-food
+3	z	inform/order	B-ingredient
+4	kurczakiem	inform/order	I-ingredient
+
+# text: proszę pizzę z szynką
+# intent: inform/order
+# slots:
+1	proszę	inform/order	NoLabel
+2	pizzę	inform/order	B-food
+3	z	inform/order	NoLabel
+4	szynką	inform/order	B-ingredient
+
+# text: poproszę pizzę barcelona w rozmiarze L
+# intent: inform/order
+# slots:
+1	poproszę	inform/order	NoLabel
+2	pizzę	inform/order	B-food
+3	barcelona	inform/order	B-pizza
+4	w	inform/order	NoLabel
+5	rozmiarze	inform/order	NoLabel
+6	L	inform/order	B-size
+
+# text: jedna pizza tuna, druga pizza hawajska
+# intent: inform/order
+# slots:
+1	jedna	inform/order	B-quantity
+2	pizza	inform/order	B-food
+3	tuna,	inform/order	B-pizza
+4	druga	inform/order	B-quantity
+5	pizza	inform/order	B-food
+6	hawajska	inform/order	B-pizza
+
+# text: wezmę pizzę capri bez pieczarek
+# intent: inform/order
+# slots:
+1	wezmę	inform/order	NoLabel
+2	pizzę	inform/order	B-food
+3	capri	inform/order	B-pizza
+4	bez	inform/order	B-ingredient/neg
+5	pieczarek	inform/order	I-ingredient/neg
+
+# text: jakie pizze są w ofercie
+# intent: request/menu
+# slots:
+1	jakie	request/menu	NoLabel
+2	pizze	request/menu	B-food
+3	są	request/menu	NoLabel
+4	w	request/menu	NoLabel
+5	ofercie	request/menu	NoLabel
+
+# text: w jakich rozmiarach są pizze
+# intent: request/size
+# slots:
+1	w	request/size	NoLabel
+2	jakich	request/size	NoLabel
+3	rozmiarach	request/size	NoLabel
+4	są	request/size	NoLabel
+5	pizze	request/size	NoLabel
+
+# text: jakie pizze są w menu
+# intent: request/menu
+# slots:
+1	jakie	request/menu	NoLabel
+2	pizze	request/menu	B-food
+3	są	request/menu	NoLabel
+4	w	request/menu	NoLabel
+5	menu	request/menu	NoLabel
+
+# text: chciałbym zamówić pizze
+# intent: request/menu
+# slots:
+1	chciałbym	request/menu	NoLabel
+2	zamówić	request/menu	NoLabel
+3	pizze	request/menu	B-food
+
+# text: co macie w ofercie
+# intent: request/menu
+# slots:
+1	co	request/menu	NoLabel
+2	macie	request/menu	NoLabel
+3	w	request/menu	NoLabel
+4	ofercie	request/menu	NoLabel
+
+# text: poproszę pizzę margarita
+# intent: inform/order
+# slots:
+1	poproszę	inform/order	NoLabel
+2	pizzę	inform/order	B-food
+3	margarita	inform/order	B-pizza
+
+# text: poproszę pizze hawajską bez ananasa
+# intent: inform/order
+# slots:
+1	poproszę	inform/order	NoLabel
+2	pizze	inform/order	B-food
+3	hawajska	inform/order	B-pizza
+4	bez	inform/order	B-ingredient/neg
+5	ananasa	inform/order	I-ingredient/neg
+
+# text: poproszę 5 pizz
+# intent: inform/order
+# slots:
+1	poproszę	inform/order	NoLabel
+2	5	inform/order	B-quantity
+3	pizz	inform/order	B-food
+
+# text: porposzę capri w rozmiarze m
+# intent: inform/order
+# slots:
+1	porposzę	inform/order	NoLabel
+2	capri	inform/order	NoLabel
+3	w	inform/order	NoLabel
+4	rozmiarze	inform/order	NoLabel
+5	m	inform/order	NoLabel
+
+# text: nie, to wszystko
+# intent: inform/order-complete
+# slots:
+1	nie,	inform/order-complete	NoLabel
+2	to	inform/order-complete	NoLabel
+3	wszystko	inform/order-complete	NoLabel
+
+# text: kiedy dojdzie dostawa?
+# intent: request/time
+# slots:
+1	kiedy	request/time	NoLabel
+2	dojdzie	request/time	NoLabel
+3	dostawa?	request/time	NoLabel
+
+# text: kiedy mogę się spodziewać dostawy?
+# intent: request/time
+# slots:
+1	kiedy	request/time	NoLabel
+2	mogę	request/time	NoLabel
+3	się	request/time	NoLabel
+4	spodziewać	request/time	NoLabel
+5	dostawy?	request/time	NoLabel
+
+# text: kiedy dojedzie kurier?
+# intent: request/time
+# slots:
+1	kiedy	request/time	NoLabel
+2	dojedzie	request/time	NoLabel
+3	kurier?	request/time	NoLabel
+
+# text: to wszystko
+# intent: inform/order-complete
+# slots:
+1	to	inform/order-complete	NoLabel
+2	wszystko	inform/order-complete	NoLabel
+
+# text: adres dostawy to Koszalińska 10/15
+# intent: inform/address
+# slots:
+1	adres	inform/address	NoLabel
+2	dostawy	inform/address	NoLabel
+3	to	inform/address	NoLabel
+4	Koszalińska	inform/address	B-address
+5	10/15	inform/address	I-address
+
+# text: proszę dzwonić na numer 320 561 123
+# intent: inform/phone
+# slots:
+1	proszę	inform/phone	NoLabel
+2	dzwonić	inform/phone	NoLabel
+3	na	inform/phone	NoLabel
+4	numer	inform/phone	NoLabel	B-phone
+5	320	inform/phone	NoLabel	I-phone
+6	561	inform/phone	NoLabel	I-phone
+7	123	inform/phone	NoLabel	I-phone
+
+# text: adres Kołątaja 15
+# intent: inform/address
+# slots:
+1	adres	inform/address	NoLabel
+2	Kołątaja	inform/address	B-address
+3	15	inform/address	I-address
+
+# text: Poznań, ul. Piastów 90/13
+# intent: inform/address
+# slots:
+1	Poznań,	inform/address	B-address
+2	ul.	inform/address	I-address
+3	Piastów	inform/address	I-address
+4	90/13	inform/address	I-address
+
+# text: ul. Piotra Skargi 13
+# intent: inform/address
+# slots:
+1	ul.	inform/address	B-address
+2	Piotra	inform/address	I-address
+3	Skargi	inform/address	I-address
+4	13	inform/address	I-address
+
+# text: aleje jerozolimskie 10
+# intent: inform
+# slots:
+1	aleje	inform/address	B-address
+2	jerozolimskie	inform/address	I-address
+3	10	inform/address	I-address
+
+# text: nie
+# intent: negate
+# slots:
+1	nie	negate	NoLabel
+
+# text: zaprzeczam
+# intent: negate
+# slots:
+1	zaprzeczam	negate	NoLabel
+
+# text: nie dziękuję
+# intent: negate
+# slots:
+1	nie	negate	NoLabel
+2	dziękuję	negate	NoLabel
+
+# text: co jest w ofercie
+# intent: request/menu
+# slots:
+1	co	request/menu	NoLabel
+2	jest	request/menu	NoLabel
+3	w	request/menu	NoLabel
+4	ofercie	request/menu	NoLabel
+
+# text: chciałbym 3 pizze, hawajskie duże
+# intent: inform/order
+# slots:
+1	chciałbym	inform/order	NoLabel
+2	3	inform/order	B-quantity
+3	pizze,	inform/order	B-food
+4	hawajskie	inform/order	B-pizza
+5	duże	inform/order	B-size
+
+# text: chciałbym barcelona xl
+# intent: inform/order
+# slots:
+1	chciałbym	inform/order	NoLabel
+2	barcelona	inform/order	B-pizza
+3	xl	inform/order	B-size
+
+# text: kruszyn krajeński, Polna 34
+# intent: inform/address
+# slots:
+1	kruszyn	inform/address	B-address
+2	krajeński,	inform/address	B-address
+3	Polna	inform/address	B-address
+4	34	inform/address	B-address
+
+# text: odbiorę na miejscu
+# intent: inform/delivery
+# slots:
+1	odbiorę	inform/delivery	NoLabel
+2	na	inform/delivery	NoLabel
+3	miejscu	inform/delivery	B-delivery
+
+# text: poproszę z dowozem
+# intent: inform/delivery
+# slots:
+1	poproszę	inform/delivery	NoLabel
+2	z	inform/delivery	NoLabel
+3	dowozem	inform/delivery	B-delivery
+
+# text: ile całość będzie kosztować
+# intent: request/price
+# slots:
+1	ile	request/price	NoLabel
+2	całość	request/price	NoLabel
+3	będzie	request/price	NoLabel
+4	kosztować	request/price	NoLabel
+
+# text: jaki koszt
+# intent: request/price
+# slots:
+1	jaki	request/price	NoLabel
+2	koszt	request/price	NoLabel
+
+# text: ile zapłacę
+# intent: request/price
+# slots:
+1	ile	request/price	NoLabel
+2	zapłacę	request/price	NoLabel
+
+# text: prosiłbym o dowóz na 17:30
+# intent: inform/time
+# slots:
+1	prosiłbym	inform/time	NoLabel
+2	o	inform/time	NoLabel
+3	dowóz	inform/time	NoLabel
+4	na	inform/time	NoLabel
+5	17:30	inform/time	B-time
+
+# text: czy można dowieźć na 20?
+# intent: inform/time
+# slots:
+1	czy	inform/time	NoLabel
+2	można	inform/time	NoLabel
+3	dowieźć	inform/time	NoLabel
+4	na	inform/time	NoLabel
+5	20?	inform/time	B-time
+
+# text: czy mozżna bez cebuli
+# intent: inform/order
+# slots:
+1	czy	inform/order	NoLabel
+2	mozżna	inform/order	NoLabel
+3	bez	inform/order	B-ingredient/neg
+4	cebuli	inform/order	I-ingredient/neg
+
+# text: bez pieczarek
+# intent: inform/order
+# slots:
+1	bez	inform/order	B-ingredient/neg
+2	pieczarek	inform/order	I-ingredient/neg
+
+# text: jaka cena?
+# intent: request/price
+# slots:
+1	jaka	request/price	NoLabel
+2	cena?	request/price	NoLabel
+
+# text: cena
+# intent: request/price
+# slots:
+1	cena	request/price	NoLabel
+
+# text: jaki koszt
+# intent: request/price
+# slots:
+1	jaki	request/price	NoLabel
+2	koszt	request/price	NoLabel
+
+# text: cena dowozu
+# intent: request/delivery-price
+# slots:
+1	cena	request/delivery-price	NoLabel
+2	dowozu	request/delivery-price	NoLabel
+
+# text: sos czosnkowy
+# intent: inform/order
+# slots:
+1	sos	inform/order	B-sauce
+2	czosnkowy	inform/order	I-sauce
+
+# text: poproszę o realizację zamówienia
+# intent: inform/order-complete
+# slots:
+1	poproszę	inform/order-complete	NoLabel
+2	o	inform/order-complete	NoLabel
+3	realizację	inform/order-complete	NoLabel
+4	zamówienia	inform/order-complete	NoLabel
+
+# text: zamawiam
+# intent: inform/order-complete
+# slots:
+1	zamawiam	inform/order-complete	NoLabel
+
+# text: i to wszystko
+# intent: inform/order-complete
+# slots:
+1	i	inform/order-complete	NoLabel
+2	to	inform/order-complete	NoLabel
+3	wszystko	inform/order-complete	NoLabel
+
+# text: jaki rozmiar
+# intent: request/size
+# slots:
+1	jaki	request/size	NoLabel
+2	rozmiar	request/size	NoLabel
+
+# text: złożyć zamówienie
+# intent: request/menu
+# slots:
+1	złożyć	request/menu	NoLabel
+2	zamówienie	request/menu	NoLabel
+
+# text: poproszę 4 pizze, tuna, hawajska, capri, barcelona, wszystkie w rozmiarze xl
+# intent: inform/order
+# slots:
+1	poproszę	inform/order	NoLabel
+2	4	inform/order	B-quantity
+3	pizze,	inform/order	B-food
+4	tuna,	inform/order	B-pizza
+5	hawajska,	inform/order	B-pizza
+6	capri,	inform/order	B-pizza
+7	barcelona,	inform/order	B-pizza
+8	wszystkie	inform/order	NoLabel
+9	w	inform/order	NoLabel
+10	rozmiarze	inform/order	NoLabel
+11	xl	inform/order	B-size
+
+# text: mój numer to 123 324 534
+# intent: inform/phone
+# slots:
+1	mój	inform/phone	NoLabel
+2	numer	inform/phone	NoLabel
+3	to	inform/phone	NoLabel
+4	123	inform/phone	B-phone
+5	324	inform/phone	I-phone
+6	534	inform/phone	I-phone
+
+# text: 653 321 787
+# intent: inform/phone
+# slots:
+1	653	inform/phone	B-phone
+2	321	inform/phone	I-phone
+3	787	inform/phone	I-phone
+
+# text: wezmę pizzę z szynką, tuńczykiem i cebulą, w rozmiarze L
+# intent: inform/order
+# slots:
+1	wezmę	inform/order	NoLabel
+2	pizzę	inform/order	B-food
+3	z	inform/order	NoLabel
+4	szynką,	inform/order	B-ingredient
+5	tuńczykiem	inform/order	B-ingredient
+6	i	inform/order	NoLabel
+7	cebulą,	inform/order	B-ingredient
+8	w	inform/order	NoLabel
+9	rozmiarze	inform/order	NoLabel
+10	L	inform/order	B-size
+
+# text: szynka
+# intent: inform/order
+# slots:
+1	szynka	inform/order	B-ingredient
+
+# text: pieczarki
+# intent: inform/order
+# slots:
+1	pieczarki	inform/order	B-ingredient
+
+# text: cebula
+# intent: inform/order
+# slots:
+1	cebula	inform/order	B-ingredient
+
+# text: papryka
+# intent: inform/order
+# slots:
+1	papryka	inform/order	B-ingredient
+
+# text: pomidor
+# intent: inform/order
+# slots:
+1	pomidor	inform/order	B-ingredient
+
+# text: ser
+# intent: inform/order
+# slots:
+1	ser	inform/order	B-ingredient
+
+# text: ananas kurczak tunczyk cebula
+# intent: inform/order
+# slots:
+1	ananas	inform/order	B-ingredient
+2	kurczak	inform/order	B-ingredient
+3	tunczyk	inform/order	B-ingredient
+4	cebula	inform/order	B-ingredient
+
+# text: barcelona hawajska tuna capri margarita
+# intent: inform/order
+# slots:
+1	barcelona	inform/order	B-pizza
+2	hawajska	inform/order	B-pizza
+3	tuna	inform/order	B-pizza
+4	capri	inform/order	B-pizza
+5	margarita	inform/order	B-pizza
+
+# text: poproszę hawajską i tuna XL
+# intent: inform/order
+# slots:
+1	poproszę	inform/order	NoLabel
+2	hawajską	inform/order	B-pizza
+3	i	inform/order	NoLabel
+4	tuna	inform/order	B-pizza
+5	XL	inform/order	B-size
+
+# text: wezmę 3 pizze tuna, średnią, dużą i bardzo dużą
+# intent: inform/order
+# slots:
+1	wezmę	inform/order	NoLabel
+2	3	inform/order	B-quantity
+3	pizze	inform/order	B-food
+4	tuna,	inform/order	B-pizza
+5	średnią,	inform/order	B-size
+6	dużą	inform/order	B-size
+7	i	inform/order	NoLabel
+8	bardzo	inform/order	B-size
+9	dużą	inform/order	I-size
+
+# text: rozmiar m
+# intent: inform/order
+# slots:
+1	rozmiar	inform/order	B-size
+2	m	inform/order	I-size
+
+# text: rozmar l
+# intent: inform/order
+# slots:
+1	rozmar	inform/order	B-size
+2	l	inform/order	I-size
+
+# text: rozmiar xl
+# intent: inform/order
+# slots:
+1	rozmiar	inform/order	B-size
+2	xl	inform/order	I-size
+
+# text: na godzine 9
+# intent: inform/time
+# slots:
+1	na	inform/time	NoLabel
+2	godzine	inform/time	B-time
+3	9	inform/time	I-time
+
+# text: godz 16
+# intent: inform/time
+# slots:
+1	godz	inform/time	B-time
+2	16	inform/time	I-time
+
+# text: na 19 poproszę
+# intent: inform/time
+# slots:
+1	na	inform/time	NoLabel
+2	19	inform/time	B-time
+3	poproszę	inform/time	NoLabel
+
+# text: godzine dziesiątą
+# intent: inform/time
+# slots:
+1	godzine	inform/time	B-time
+2	dziesiątą	inform/time	I-time
+
+# text: na szóstą
+# intent: inform/time
+# slots:
+1	na	inform/time	NoLabel
+2	szóstą	inform/time	B-time
+
+# text: tak zgadza się
+# intent: affirm
+# slots:
+1	tak	affirm	NoLabel
+2	zgadza	affirm	NoLabel
+3	się	affirm	NoLabel
+
+# text: potwierdzam
+# intent: affirm
+# slots:
+1	potwierdzam	affirm	NoLabel
+
+# text: nie zgadza się
+# intent: negate
+# slots:
+1	nie	negate	NoLabel
+2	zgadza	negate	NoLabel
+3	się	negate	NoLabel
+
+# text: numer to 333444555
+# intent: inform/phone
+# slots:
+1	numer	inform/phone	NoLabel
+2	to	inform/phone	NoLabel
+3	333444555	inform/phone	B-phone
+
+# text: numer to 335 23 25 29
+# intent: inform/phone
+# slots:
+1	numer	inform/phone	NoLabel
+2	to	inform/phone	NoLabel
+3	335	inform/phone	B-phone
+4	23	inform/phone	I-phone
+5	25	inform/phone	I-phone
+6	29	inform/phone	I-phone
+
+# text: dziękuję to wszystko
+# intent: inform/order-complete
+# slots:
+1	dziękuję	inform/order-complete	NoLabel
+2	to	inform/order-complete	NoLabel
+3	wszystko	inform/order-complete	NoLabel
+
+# text: potwierdzam zamówienie
+# intent: inform/order-complete
+# slots:
+1	potwierdzam	inform/order-complete	NoLabel
+2	zamówienie	inform/order-complete	NoLabel
+
+# text: bez kurczaka
+# intent: inform/order
+# slots:
+1	bez	inform/order	B-ingredient/neg
+2	kurczaka	inform/order	I-ingredient/neg
+
+# text: bez cebuli
+# intent: inform/order
+# slots:
+1	bez	inform/order	B-ingredient/neg
+2	cebuli	inform/order	I-ingredient/neg
+
+# text: bez szynki i pieczarek
+# intent: inform/order
+# slots:
+1	bez	inform/order	B-ingridient/neg
+2	szynki	inform/order	I-ingredient/neg
+3	i	inform/order	NoLabel
+4	pieczarek	inform/order	B-ingredient/neg
+
+# text: proszę pizzę z ananasem, kurczakiem i serem
+# intent: inform/order
+# slots:
+1	proszę	inform/order	NoLabel
+2	pizzę	inform/order	B-food
+3	z	inform/order	NoLabel
+4	ananasem,	inform/order	B-ingredient
+5	kurczakiem	inform/order	B-ingredient
+6	i	inform/order	NoLabel
+7	serem	inform/order	B-ingredient
+
+# text: ile płacę
+# intent: request/price
+# slots:
+1	ile	request/price	NoLabel
+2	płacę	request/price	NoLabel
+
+# text: ile zapłacę
+# intent: request/price
+# slots:
+1	ile	request/price	NoLabel
+2	zapłacę	request/price	NoLabel
+
+# text: sos tysiąca wysp
+# intent: inform/order
+# slots:
+1	sos	inform/order	B-sauce
+2	tysiąca	inform/order	I-sauce
+3	wysp	inform/order	I-sauce
+
+# text: sos koperkowy
+# intent: inform/order
+# slots:
+1	sos	inform/order	B-sauce
+2	koperkowy	inform/order	I-sauce
+
+# text: tak dziękuję
+# intent: affirm
+# slots:
+1	tak	affirm	NoLabel
+2	dziękuję	affirm	NoLabel
+
+# text: nnie, dziękuję
+# intent: negate
+# slots:
+1	nnie,	negate	NoLabel
+2	dziękuję	negate	NoLabel
+
+# text: nie dzięki
+# intent: negate
+# slots:
+1	nie	negate	NoLabel
+2	dzięki	negate	NoLabel

evaluate.py

@@ -0,0 +1,45 @@
+import re
+import os
+import pandas as pd
+import numpy as np
+from nlu_utils import predict_multiple
+from flair.models import SequenceTagger
+
+def __parse_acts(acts):
+    acts_split = acts.split('&')
+    remove_slot_regex = "[\(\[].*?[\)\]]"
+    return set(re.sub(remove_slot_regex, "", act) for act in acts_split)
+
+def __parse_predictions(predictions):
+    return set(prediction.split('/')[0] for prediction in predictions)
+
+# Exploratory tests
+frame_model = SequenceTagger.load('frame-model-prod/best-model.pt')
+# slot_model = SequenceTagger.load('slot-model-prod/final-model.pt')
+
+total_acts = 0
+act_correct_predictions = 0
+slot_correct_predictions = 0
+
+for file_name in os.listdir('data'):
+    if file_name.split('.')[-1] != 'tsv':
+        continue
+
+    df = pd.read_csv(f'data/{file_name}', sep='\t', names=['kto', 'treść', 'akt'])
+    df = df[df.kto == 'user']
+    all_data = np.array(df)
+
+    for row in all_data:
+        sentence = row[1]
+        acts = __parse_acts(row[2])
+
+        predictions_raw = predict_multiple(frame_model, sentence.split(), 'frame')
+        predictions = __parse_predictions(predictions_raw)
+
+        for act in acts:
+            total_acts += 1
+            if act in predictions:
+                act_correct_predictions += 1
+                
+
+print(f"Accuracy - predicting acts: {(act_correct_predictions / total_acts)*100} ({act_correct_predictions}/{total_acts})")

nlu_tests.py

@@ -0,0 +1,30 @@
+from flair.models import SequenceTagger
+from nlu_utils import predict_single, predict_multiple, predict_and_annotate
+
+# Exploratory tests
+frame_model = SequenceTagger.load('frame-model/best-model.pt')
+tests = [
+    'chciałbym zamówić pizzę',
+    'na godzinę 12',
+    'prosiłbym o pizzę z pieczarkami',
+    'to wszystko, jaka cena?',
+    'ile kosztuje pizza',
+    'do widzenia',
+    'tak',
+    'nie dziękuję',
+    'dodatkowy ser',
+    'pizzę barcelona bez cebuli',
+]
+
+# print("=== Exploratory tests - frame model ===")
+for test in tests:
+    print(f"Sentence: {test}")
+    print(f"Single prediction: {predict_single(frame_model, test.split(), 'frame')}")
+    print(f"Multiple predictions: {predict_multiple(frame_model, test.split(), 'frame')}")
+    print(f"Annotated sentence: {predict_and_annotate(frame_model, test.split(), 'frame')}")
+
+print("=== Exploratory tests - slot model ===")
+slot_model = SequenceTagger.load('slot-model/final-model.pt')
+for test in tests:
+    print(f"Sentence: {test}")
+    print(f"Prediction: {predict_and_annotate(slot_model, test.split(), 'slot')}")

nlu_train.py

@@ -0,0 +1,46 @@
+from conllu import parse_incr
+from flair.data import Corpus
+from flair.embeddings import StackedEmbeddings
+from flair.embeddings import WordEmbeddings
+from flair.embeddings import CharacterEmbeddings
+from flair.embeddings import FlairEmbeddings
+from flair.models import SequenceTagger
+from flair.trainers import ModelTrainer
+from nlu_utils import conllu2flair, nolabel2o
+
+import random
+import torch
+random.seed(42)
+torch.manual_seed(42)
+
+if torch.cuda.is_available():
+    torch.cuda.manual_seed(0)
+    torch.cuda.manual_seed_all(0)
+    torch.backends.cudnn.enabled = False
+    torch.backends.cudnn.benchmark = False
+    torch.backends.cudnn.deterministic = True
+
+
+def train_model(label_type, field_parsers = {}):
+    with open('data/train_dialog.conllu', encoding='utf-8') as trainfile:
+        trainset = list(parse_incr(trainfile, fields=['id', 'form', 'frame', 'slot'], field_parsers=field_parsers))
+
+    corpus = Corpus(train=conllu2flair(trainset, label_type), test=conllu2flair(trainset, label_type))
+    label_dictionary = corpus.make_label_dictionary(label_type=label_type)
+
+    embedding_types = [
+        WordEmbeddings('pl'),
+        FlairEmbeddings('pl-forward'),
+        FlairEmbeddings('pl-backward'),
+        CharacterEmbeddings(),
+    ]
+
+    embeddings = StackedEmbeddings(embeddings=embedding_types)
+    tagger = SequenceTagger(hidden_size=256, embeddings=embeddings, tag_dictionary=label_dictionary, tag_type=label_type, use_crf=True, tag_format="BIO")
+                
+    frame_trainer = ModelTrainer(tagger, corpus)
+    frame_trainer.train(f'{label_type}-model', learning_rate=0.1, mini_batch_size=32, max_epochs=75, train_with_dev=False)
+
+if __name__ == '__main__':
+    train_model("frame")
+    train_model('slot', field_parsers={'slot': nolabel2o})

nlu_utils.py

@@ -0,0 +1,100 @@
+from flair.data import Sentence
+from flair.datasets import FlairDatapointDataset
+
+def nolabel2o(line, i):
+    return 'O' if line[i] == 'NoLabel' else line[i]
+
+def conllu2flair(sentences, label=None):
+    if label == "frame":
+        return conllu2flair_frame(sentences, label)
+    else:
+        return conllu2flair_slot(sentences, label)
+
+def conllu2flair_frame(sentences, label=None):
+    fsentences = []
+    for sentence in sentences:
+        tokens = [token["form"] for token in sentence]
+        fsentence = Sentence(' '.join(tokens), use_tokenizer=False)
+
+        for i in range(len(fsentence)):
+            fsentence[i:i+1].add_label(label, sentence[i][label])
+
+        fsentences.append(fsentence)
+
+    return FlairDatapointDataset(fsentences)
+
+def conllu2flair_slot(sentences, label=None):
+    fsentences = []
+
+    for sentence in sentences:
+        fsentence = Sentence(' '.join(token['form'] for token in sentence), use_tokenizer=False)
+        start_idx = None
+        end_idx = None
+        tag = None
+
+        if label:
+            for idx, (token, ftoken) in enumerate(zip(sentence, fsentence)):
+                if token[label].startswith('B-'):
+                    start_idx = idx
+                    end_idx = idx
+                    tag = token[label][2:]
+                elif token[label].startswith('I-'):
+                    end_idx = idx
+                elif token[label] == 'O':
+                    if start_idx is not None:
+                        fsentence[start_idx:end_idx+1].add_label(label, tag)
+                        start_idx = None
+                        end_idx = None
+                        tag = None
+
+            if start_idx is not None:
+                fsentence[start_idx:end_idx+1].add_label(label, tag)
+
+        fsentences.append(fsentence)
+    return FlairDatapointDataset(fsentences)
+
+def __predict(model, csentence):
+    fsentence = conllu2flair([csentence])[0]
+    model.predict(fsentence)
+    return fsentence
+
+def __csentence(sentence, label_type):
+    if label_type == "frame":
+        return [{'form': word } for word in sentence]
+    else:
+        return [{'form': word, 'slot': 'O'} for word in sentence]
+
+def predict_single(model, sentence, label_type):
+    csentence = __csentence(sentence, label_type)
+    fsentence = __predict(model, csentence)
+    intent = {}
+
+    for span in fsentence.get_spans(label_type):
+        tag = span.get_label(label_type).value
+        if tag in intent:
+            intent[tag] += 1
+        else:
+            intent[tag] = 1
+
+    return max(intent, key=intent.get)
+
+def predict_multiple(model, sentence, label_type):
+    csentence = __csentence(sentence, label_type)
+    fsentence = __predict(model, csentence)
+
+    return set(span.get_label(label_type).value for span in fsentence.get_spans(label_type))
+
+def predict_and_annotate(model, sentence, label_type):
+    csentence = __csentence(sentence, label_type)
+    fsentence = __predict(model, csentence)
+
+    for span in fsentence.get_spans(label_type):
+        tag = span.get_label(label_type).value
+        if label_type == "frame":
+            csentence[span.tokens[0].idx-1]['frame'] = tag
+        else:
+            csentence[span.tokens[0].idx - 1]['slot'] = f'B-{tag}'
+            for token in span.tokens[1:]:
+                csentence[token.idx - 1]['slot'] = f'I-{tag}'
+
+    return csentence