Compare commits

...

2 Commits

Author SHA1 Message Date
71f525ec07 Change requirements 2024-06-09 15:06:51 +02:00
312225b307 dialog_policy 2024-06-07 14:45:13 +02:00
14 changed files with 254 additions and 9 deletions

View File

@ -1,5 +1,6 @@
flair==0.13.1
conllu==4.5.3
pandas==1.5.3
numpy==1.26.4
numpy==1.24.4
torch==2.3.0
fuzzywuzzy==0.18.0

View File

@ -0,0 +1,24 @@
"""Policy Interface"""
from convlab.util.module import Module
class Policy(Module):
"""Policy module interface."""
def predict(self, state):
"""Predict the next agent action given dialog state.
Args:
state (dict or list of list):
when the policy takes dialogue state as input, the type is dict.
else when the policy takes dialogue act as input, the type is list of list.
Returns:
action (list of list or str):
when the policy outputs dialogue act, the type is list of list.
else when the policy outputs utterance directly, the type is str.
"""
return []
def update_memory(self, utterance_list, state_list, action_list, reward_list):
pass

View File

@ -0,0 +1,25 @@
"""module interface."""
from abc import ABC
class Module(ABC):
def train(self, *args, **kwargs):
"""Model training entry point"""
pass
def test(self, *args, **kwargs):
"""Model testing entry point"""
pass
def from_cache(self, *args, **kwargs):
"""restore internal state for multi-turn dialog"""
return None
def to_cache(self, *args, **kwargs):
"""save internal state for multi-turn dialog"""
return None
def init_session(self):
"""Init the class variables for a new session."""
pass

View File

@ -0,0 +1,38 @@
"""
"""
import json
import os
import random
from fuzzywuzzy import fuzz
from itertools import chain
from copy import deepcopy
class Database(object):
def __init__(self):
super(Database, self).__init__()
# loading databases
domains = ['menu', 'pizza', 'drink', 'size']
self.dbs = {}
for domain in domains:
with open(os.path.join(os.path.dirname(
os.path.dirname(os.path.dirname(os.path.dirname(os.path.abspath(__file__))))),
'data/restaurant/db/{}_db.json'.format(domain))) as f:
self.dbs[domain] = json.load(f)
def query(self, domain):
"""Returns the list of entities for a given domain
based on the annotation of the belief state"""
# query the db
if domain == 'pizza':
return [{'Name': random.choice(self.dbs[domain]['name'])}]
if domain == 'menu':
return deepcopy(self.dbs[domain])
if domain == 'drink':
return [{'Name': random.choice(self.dbs[domain]['name'])}]
if domain == 'size':
return [{'Size': random.choice(self.dbs[domain]['size'])}]
if __name__ == '__main__':
db = Database()

View File

@ -0,0 +1,4 @@
[
"true",
"false"
]

View File

@ -0,0 +1,5 @@
[
"pepsi",
"cola",
"water"
]

View File

@ -0,0 +1,11 @@
[
{
"name":"pepsi"
},
{
"name":"cola"
},
{
"name":"water"
}
]

View File

@ -0,0 +1,3 @@
[
"pizza"
]

View File

@ -0,0 +1,5 @@
[
"chicken",
"ham",
"tuna"
]

View File

@ -0,0 +1,7 @@
[
"capri",
"margarita",
"hawajska",
"barcelona",
"tuna"
]

View File

@ -0,0 +1,51 @@
[
{
"name": "capri",
"ingredient": [
"tomato",
"ham",
"mushrooms",
"cheese"
],
"price": 25
},
{
"name": "margarita",
"ingredient": [
"tomato",
"cheese"
],
"price": 20
},
{
"name": "hawajska",
"ingredient": [
"tomato",
"pineapple",
"chicken",
"cheese"
],
"price": 30
},
{
"name": "barcelona",
"ingredient": [
"tomato",
"onion",
"ham",
"pepper",
"cheese"
],
"price": 40
},
{
"name": "tuna",
"ingredient": [
"tomato",
"tuna",
"onion",
"cheese"
],
"price": 40
}
]

View File

@ -0,0 +1,4 @@
[
"garlic",
"1000w"
]

View File

@ -0,0 +1,14 @@
[
{
"size": "m",
"price_multiplier": 1
},
{
"size": "l",
"price_multiplier": 1.2
},
{
"size": "xl",
"price_multiplier": 1.4
}
]

View File

@ -1,8 +1,61 @@
from model.frame import Frame
from collections import defaultdict
import copy
import json
from copy import deepcopy
class DialogPolicy:
def next_dialogue_act(self, frames: list[Frame]) -> Frame:
if frames[-1].act == "welcomemsg":
return Frame("system", "welcomemsg", [])
else:
return Frame("system", "canthelp", [])
from convlab.policy.policy import Policy
from convlab.util.restaurant.dbquery import Database
class SimpleRulePolicy(Policy):
def __init__(self):
Policy.__init__(self)
self.db = Database()
def predict(self, state):
self.results = []
system_action = defaultdict(list)
user_action = defaultdict(list)
for intent, domain, slot, value in state['user_action']:
user_action[(domain.lower(), intent.lower())].append((slot.lower(), value))
for user_act in user_action:
self.update_system_action(user_act, user_action, state, system_action)
# Reguła 3
if any(True for slots in user_action.values() for (slot, _) in slots if slot in ['pizza', 'size', 'drink']):
if self.results:
system_action = {('Ordering', 'Order'): [["Ref", self.results[0].get('Ref', 'N/A')]]}
system_acts = [[intent, domain, slot, value] for (domain, intent), slots in system_action.items() for slot, value in slots]
state['system_action'] = system_acts
return system_acts
def update_system_action(self, user_act, user_action, state, system_action):
domain, intent = user_act
constraints = [(slot, value) for slot, value in state['belief_state'][domain.lower()].items() if value != '']
self.results = deepcopy(self.db.query(domain.lower(), constraints))
# Reguła 1
if intent == 'request':
if len(self.results) == 0:
system_action[(domain, 'NoOffer')] = []
else:
for slot in user_action[user_act]:
if slot[0] in self.results[0]:
system_action[(domain, 'Inform')].append([slot[0], self.results[0].get(slot[0], 'unknown')])
# Reguła 2
elif intent == 'inform':
if len(self.results) == 0:
system_action[(domain, 'NoOffer')] = []
else:
system_action[(domain, 'Inform')].append(['Choice', str(len(self.results))])
choice = self.results[0]
if domain in ["pizza", "drink"]:
system_action[(domain, 'Recommend')].append(['Name', choice['name']])
if domain in ["size"]:
system_action[(domain, 'Recommend')].append(['Size', choice['size']])
dialogPolicy = SimpleRulePolicy()