def addAutobiographicMemory(threshold = 1.5):
# memory:
# - time tag
# - delta mood vector
# - context
# - recent cmd hist
new_mem = {}
memLen = 3
# check that memory is of sufficient emotional magnitude
# deltaM12 = delta mood
cNum = len(GLOB['CMDHIST'])
if cNum < memLen:
return False
oldMood = GLOB['CMDHIST'][cNum-2][3]
newMood = GLOB['CMDHIST'][cNum-1][3]
deltaMood = [round(newMood[i] - oldMood[i], 2) for i in range(GLOB['MOODDIM'])]
maxDelta = max([abs(item) for item in deltaMood])
memoryWeight = sum([abs(item) for item in deltaMood])
NowTime = datetime.datetime.now()
NowTotalSecs = (NowTime-datetime.datetime(1970,1,1)).total_seconds()
new_mem['deltaMood'] = deltaMood
#new_mem['mood'] = GLOB['CMDHIST'][cNum-1][3]
new_mem['weight'] =round(memoryWeight, 3)
new_mem['time'] = NowTotalSecs
new_mem['context'] = GLOB['CONTEXT']
# get the last memLen commands: [old, .. , new]
new_mem['recentHist'] = [GLOB['CMDHIST'][cNum-1-i][1] for i in range(memLen-1, -1, -1)]
new_mem['novelty'] = round(getMemoryNovelty(new_mem), 3)
new_mem['title'] = getMemoryTitle(new_mem)
new_mem['count_recalled'] = 0
new_mem['count_happened'] = 0
#print "weight, maxDelta, novelty = ", memoryWeight, maxDelta, new_mem['novelty']
if new_mem['novelty'] >= threshold and (new_mem['weight'] >= 1.5 or maxDelta >= 0.5):
GLOB['AUTOBIOGRAPHIC_MEMORY'].append(new_mem)
jsonSave(GLOB['AUTOBIOGRAPHIC_MEMORY'], DATA_DIR+'autobiographic_memory.txt')
playMemSound(new_mem)
return True
else:
return False
def saveData(self, event=[]):
#print "saving..."
pixelX=self.root.winfo_width()
pixelY=self.root.winfo_height()
oX = self.root.winfo_rootx()
oY = self.root.winfo_rooty()
data = {}
data['root_geometry'] = self.root.winfo_geometry() #[pixelX, pixelY, oX, oY]
#"1097x499+94+212"
data['thoughts'] = []
data['zoom'] = self.curZoom
for t in self.thoughts:
tData={}
#tData['index'] = t.index
tData['pixLoc'] = t.pixLoc
tData['radius'] = t.r
tData['text'] = t.getText()
tData['fontSize'] = t.fontSize
data['thoughts'].append(tData)
data['links'] = []
for l in self.links:
lData={}
#need to add 1 since index assignments for thoughts starts at 1
# instead of 0
lData['tA'] = self.thoughts.index(l.tA)+1
lData['tB'] = self.thoughts.index(l.tB)+1
lData['importance'] = l.importance
data['links'].append(lData)
#l.grow()
fileIO.jsonSave(data=data, fileName=self.fileName, indent=True, sort=False, oneLine=False)
self.pulse()
return
def rememberMemory(mem, confidence, memType):
from emotions import ingest
if mem == {} or memType == "":
return
if (confidence < 0.4 and memType == "remember"):
return False
elif (confidence < 0.6 and memType == "predict"):
return False
playMemSound(mem)
importance = memoryImportance(mem, confidence)
# UPDATE MEMORY DETAILS
memIndex = GLOB['AUTOBIOGRAPHIC_MEMORY'].index(mem)
GLOB['AUTOBIOGRAPHIC_MEMORY'][memIndex]['count_recalled'] += 1
if confidence >= 0.95 and memType == "remember":
GLOB['AUTOBIOGRAPHIC_MEMORY'][memIndex]['count_happened'] += 1
jsonSave(GLOB['AUTOBIOGRAPHIC_MEMORY'],
DATA_DIR + 'autobiographic_memory.txt')
# decide if we're going to act on the memory
# count_recalled
# count_happened
# -> if the memory is often recalled but never happens and weight is low
# then we don't act on it
if memoryImportance(mem, confidence) < 0.5:
# print "IMPORTANCE = ", importance
return False
#sayStr = "*I "+getMemoryStr(mem, confidence, memType)+' %s'%round(importance, 2)+'.*'
#sayStr += ' '+mem['title']
sayStr = "*I " + getMemoryStr(mem, confidence, memType) + '.*'
say(sayStr, more=False, speak=setting("SPEAK"), location="top")
if sum(mem['deltaMood']) >= 0:
memoryAnimation(mood="positive")
else:
memoryAnimation(mood="negative")
memMood = [m[0] + m[1] for m in zip(GLOB['MOOD'], mem['deltaMood'])]
memMood = [max(min(memMood[i], 1), -1) for i in range(GLOB['MOODDIM'])]
# weight of each facet of memory is equal to the magnitude of the
# emotion change in that memory
memWeight = [abs(m) * confidence for m in mem['deltaMood']]
if memType == "predict":
memWeight = [m * 0.25 for m in memWeight]
# add memory effects to mood
ingest(inCmd=[],
inStr=mem['title'],
mood=memMood,
weight=memWeight,
isMem=True)
return True
def addAutobiographicMemory(threshold=1.5):
# memory:
# - time tag
# - delta mood vector
# - context
# - recent cmd hist
new_mem = {}
memLen = 3
# check that memory is of sufficient emotional magnitude
# deltaM12 = delta mood
cNum = len(GLOB['CMDHIST'])
if cNum < memLen:
return False
oldMood = GLOB['CMDHIST'][cNum - 2][3]
newMood = GLOB['CMDHIST'][cNum - 1][3]
deltaMood = [
round(newMood[i] - oldMood[i], 2) for i in range(GLOB['MOODDIM'])
]
maxDelta = max([abs(item) for item in deltaMood])
memoryWeight = sum([abs(item) for item in deltaMood])
NowTime = datetime.datetime.now()
NowTotalSecs = (NowTime - datetime.datetime(1970, 1, 1)).total_seconds()
new_mem['deltaMood'] = deltaMood
#new_mem['mood'] = GLOB['CMDHIST'][cNum-1][3]
new_mem['weight'] = round(memoryWeight, 3)
new_mem['time'] = NowTotalSecs
new_mem['context'] = GLOB['CONTEXT']
# get the last memLen commands: [old, .. , new]
new_mem['recentHist'] = [
GLOB['CMDHIST'][cNum - 1 - i][1] for i in range(memLen - 1, -1, -1)
]
new_mem['novelty'] = round(getMemoryNovelty(new_mem), 3)
new_mem['title'] = getMemoryTitle(new_mem)
new_mem['count_recalled'] = 0
new_mem['count_happened'] = 0
#print "weight, maxDelta, novelty = ", memoryWeight, maxDelta, new_mem['novelty']
if new_mem['novelty'] >= threshold and (new_mem['weight'] >= 1.5
or maxDelta >= 0.5):
GLOB['AUTOBIOGRAPHIC_MEMORY'].append(new_mem)
jsonSave(GLOB['AUTOBIOGRAPHIC_MEMORY'],
DATA_DIR + 'autobiographic_memory.txt')
playMemSound(new_mem)
return True
else:
return False
def rememberMemory(mem, confidence, memType):
from emotions import ingest
if mem=={} or memType=="":
return
if (confidence < 0.4 and memType == "remember"):
return False
elif (confidence < 0.6 and memType == "predict"):
return False
playMemSound(mem)
importance = memoryImportance(mem, confidence)
# UPDATE MEMORY DETAILS
memIndex = GLOB['AUTOBIOGRAPHIC_MEMORY'].index(mem)
GLOB['AUTOBIOGRAPHIC_MEMORY'][memIndex]['count_recalled'] += 1
if confidence >= 0.95 and memType == "remember":
GLOB['AUTOBIOGRAPHIC_MEMORY'][memIndex]['count_happened'] += 1
jsonSave(GLOB['AUTOBIOGRAPHIC_MEMORY'], DATA_DIR+'autobiographic_memory.txt')
# decide if we're going to act on the memory
# count_recalled
# count_happened
# -> if the memory is often recalled but never happens and weight is low
# then we don't act on it
if memoryImportance(mem, confidence) < 0.5:
# print "IMPORTANCE = ", importance
return False
#sayStr = "*I "+getMemoryStr(mem, confidence, memType)+' %s'%round(importance, 2)+'.*'
#sayStr += ' '+mem['title']
sayStr = "*I "+getMemoryStr(mem, confidence, memType)+'.*'
say(sayStr, more=False, speak=setting("SPEAK"), location="top")
if sum(mem['deltaMood']) >= 0:
memoryAnimation(mood="positive")
else:
memoryAnimation(mood="negative")
memMood=[m[0]+m[1] for m in zip(GLOB['MOOD'], mem['deltaMood'])]
memMood = [max(min(memMood[i], 1), -1) for i in range(GLOB['MOODDIM'])]
# weight of each facet of memory is equal to the magnitude of the
# emotion change in that memory
memWeight = [abs(m)*confidence for m in mem['deltaMood']]
if memType == "predict":
memWeight = [m*0.25 for m in memWeight]
# add memory effects to mood
ingest(inCmd=[], inStr=mem['title'], mood=memMood, weight=memWeight, isMem=True)
return True
def handleInput(inStr, factList, factFileName, getContext=False, queries=True, save=False):
global W5
inStr = removeStrange(inStr).strip()
#print "HANDLING:", inStr
if "tag:" == inStr.split()[0]:
tagStr = inStr.lower().replace('tag: ', '')
retStr = formAnswerString(tagQuery(tag=tagStr, factList=factList))
return retStr, tagStr, "add"
topMatches = []
topWType = []
topConfidence = 0
inputType = "add"
for wK in W5:
matches_add, confidence_add = matchFromList(W5[wK]['add'], inStr, trials=10, subTrials=150, includeLenScore=True)
matches_query, confidence_query = matchFromList(W5[wK]['query'], inStr, trials=10, subTrials=150, includeLenScore=True)
if DEBUG:
print "clearlyQuestion:", clearlyQuestion(inStr)
print matches_add, confidence_add
print matches_query, confidence_query
if confidence_add >= topConfidence and not clearlyQuestion(inStr):
topConfidence = confidence_add
topMatches = matches_add
topWType = wK
inputType = "add"
#print matches_add, confidence_add
elif confidence_query > topConfidence:
topConfidence = confidence_query
topMatches = matches_query
topWType = wK
inputType = "query"
#print matches_query, confidence_query
if topMatches == [] or (inputType=="add" and len(topMatches[0]) < 2) or topConfidence < 0.95:
# no good match found, so return default
#print "here"
#print topConfidence
return "", "", ""
retStr = ""
context=""
dataChanged = False
if inputType == "add":
# ex. what(sub) = prob
sub = topMatches[0][0]
prop = topMatches[0][1]
replaceInfo = "just" in topMatches[1] or "only" in topMatches[1]
retStr="Adding to factual memory:\n"+'['+topWType+']'+'('+sub+') = '+prop
fact = [sub, topWType, prop] # ex. ["apples", "what who", "red"]
addToFacts(factList, fact, replaceInfo=replaceInfo)
dataChanged = True
context = prop
elif inputType == "query" and queries:
answers=queryFacts(factList=factList, query=inStr, qType=topWType, matches=topMatches, minScore=0.3)
#print "ANSWERS:", answers
retStr = formAnswerString(answers)#, qType=topWType)
if len(answers[0]) != 0:
context = answers[0][2]
if save and dataChanged:
jsonSave(factList, factFileName)
return retStr, context, inputType
W5 = jsonLoad('w5_4.txt')
factList = jsonLoad(factFileName)
DEBUG =True
inStr = ""
while inStr != "exit":
sys.stdout.write('>> ')
inStr = raw_input("")
print handleInput(inStr=inStr, factList=factList, factFileName=factFileName, save=True)
jsonSave(factList, factFileName)
'''
file = open('raw_text.txt')
t=file.read()
TB = TextBlob(t)
TS = [item.raw for item in TB.sentences]
for i in range(len(TS)):
s = TS[i]
print i+1, "/", len(TS)
handleInput(inStr=s, factList=factList, factFileName=factFileName)
if i >= 50: break
