copycat/formulas.py

import math #, random
import logging

import utils

from temperature import temperature

actualTemperature = Temperature = 100.0

def selectListPosition(probabilities):
	total = sum(probabilities)
	#logging.info('total: %s' % total)
	r = utils.random()
	stopPosition = total * r
	#logging.info('stopPosition: %s' % stopPosition)
	total = 0
	index = 0
	for probability in probabilities:
		total += probability
		if total > stopPosition:
			return index
		index += 1
	return 0

def weightedAverage(values):
	total = 0.0
	totalWeights = 0.0
	for value,weight in values:
		total += value * weight
		totalWeights += weight
	if not totalWeights:
		return 0.0
	return total / totalWeights

def temperatureAdjustedValue(value):
	#logging.info('Temperature: %s' % Temperature)
	#logging.info('actualTemperature: %s' % actualTemperature)
	return value ** (((100.0-Temperature)/30.0)+0.5)

def temperatureAdjustedProbability(value):
	if not value or value == 0.5 or not temperature.value:
		return value
	if value < 0.5:
		return 1.0 - temperatureAdjustedProbability(1.0 - value)
	coldness = 100.0 - temperature.value
	a = math.sqrt(coldness)
	b = 10.0 - a
	c = b / 100
	d = c * ( 1.0 - ( 1.0 - value ) ) # aka c * value, but we're following the java
	e = ( 1.0 - value ) + d
	f = 1.0 - e
	return max(f,0.5)

def coinFlip(chance=0.5):
	return utils.random() < chance

def blur(value):
	root = math.sqrt(value)
	if coinFlip():
		return value + root
	return value - root

def chooseObjectFromList(objects,attribute):
	if not objects:
		return None
	probabilities = []
	for object in objects:
		value = getattr(object,attribute)
		probability = temperatureAdjustedValue(value)
		logging.info('Object: %s, value: %d, probability: %d' % (object,value,probability))
		probabilities += [ probability ]
	index = selectListPosition(probabilities)
	logging.info("Selected: %d" % index)
	return objects[index]

def chooseRelevantDescriptionByActivation(workspaceObject):
	descriptions = workspaceObject.relevantDescriptions()
	if not descriptions:
		return None
	activations = [ description.descriptor.activation for description in descriptions ]
	index = selectListPosition(activations)
	return descriptions[ index ]

def similarPropertyLinks(slip_node):
	result = []
	for slip_link in slip_node.propertyLinks:
		association = slip_link.degreeOfAssociation() / 100.0
		probability = temperatureAdjustedProbability( association )
		if coinFlip(probability):
			result += [ slip_link ]
	return result

def chooseSlipnodeByConceptualDepth(slip_nodes):
	if not slip_nodes:
		return None
	depths = [ temperatureAdjustedValue(n.conceptualDepth) for n in slip_nodes ]
	i = selectListPosition(depths)
	return slip_nodes[ i ]

def __relevantCategory(objekt,slipnode):
	return objekt.rightBond and objekt.rightBond.category == slipnode

def __relevantDirection(objekt,slipnode):
	return objekt.rightBond and objekt.rightBond.directionCategory == slipnode

def __localRelevance(string,slipnode,relevance):
	numberOfObjectsNotSpanning = numberOfMatches = 0.0
	#logging.info("find relevance for a string: %s" % string);
	for objekt in string.objects:
		#logging.info('object: %s' % objekt)
		if not objekt.spansString():
			#logging.info('non spanner: %s' % objekt)
			numberOfObjectsNotSpanning += 1.0
			if relevance(objekt,slipnode):
				numberOfMatches += 1.0
	#logging.info("matches: %d, not spanning: %d" % (numberOfMatches,numberOfObjectsNotSpanning))
	if numberOfObjectsNotSpanning == 1:
		return 100.0 * numberOfMatches
	return 100.0 * numberOfMatches / (numberOfObjectsNotSpanning - 1.0)

def localBondCategoryRelevance(string, category):
	if len(string.objects) == 1:
		return 0.0
	return __localRelevance(string,category,__relevantCategory)

def localDirectionCategoryRelevance(string, direction):
	return __localRelevance(string,direction,__relevantDirection)

def getMappings(objectFromInitial,objectFromTarget, initialDescriptions, targetDescriptions):
	mappings = []
	from conceptMapping import ConceptMapping
	for initialDescription in initialDescriptions:
		for targetDescription in targetDescriptions:
			if initialDescription.descriptionType == targetDescription.descriptionType:
				if initialDescription.descriptor == targetDescription.descriptor or initialDescription.descriptor.slipLinked(targetDescription.descriptor):
					mapping = ConceptMapping(
						initialDescription.descriptionType,
						targetDescription.descriptionType,
						initialDescription.descriptor,
						targetDescription.descriptor,
						objectFromInitial,
						objectFromTarget
					)
					mappings += [ mapping ]
	return mappings