def __init__(self):
ToolChain.__init__(self)
# Steps
self.addStep("CONVERT", self.convert, {"dataSetNames":None, "corpusName":None} , "documents.xml")
self.addStep("SPLIT-SENTENCES", Tools.GeniaSentenceSplitter.makeSentences, {"debug":False, "postProcess":True}, "sentences.xml")
self.addStep("NER", Tools.BANNER.run, {"elementName":"entity", "processElement":"sentence", "debug":False, "splitNewlines":True}, "ner.xml")
self.addStep("PARSE", Tools.CharniakJohnsonParser.parse, {"parseName":"McCC", "requireEntities":False, "debug":False}, "parse.xml")
self.addStep("CONVERT-PARSE", Tools.StanfordParser.convertXML, {"parser":"McCC", "debug":False}, "converted-parse.xml")
self.addStep("SPLIT-NAMES", ProteinNameSplitter.mainFunc, {"parseName":"McCC"}, "split-names.xml")
self.addStep("FIND-HEADS", FindHeads.findHeads, {"parse":"McCC", "removeExisting":True}, "heads.xml")
self.addStep("DIVIDE-SETS", self.divideSets, {"outputStem":None, "saveCombined":True})
def __init__(self, steps, parseName="McCC", requireEntities=False):
#if constParser == "None": constParser = None
#if depParser == "None": depParser = None
#assert constParser in ("BLLIP", "BLLIP-BIO", "STANFORD", None), constParser
#assert depParser in ("STANFORD", "STANFORD-CONVERT", "SYNTAXNET", None), depParser
#self.constParser = constParser
#self.depParser = depParser
self.requireEntities = requireEntities
self.parseName = parseName
ToolChain.__init__(self, steps)
self.modelParameterStringName = "preprocessorParams"
def __init__(self, steps, parseName="McCC", requireEntities=False):
#if constParser == "None": constParser = None
#if depParser == "None": depParser = None
#assert constParser in ("BLLIP", "BLLIP-BIO", "STANFORD", None), constParser
#assert depParser in ("STANFORD", "STANFORD-CONVERT", "SYNTAXNET", None), depParser
#self.constParser = constParser
#self.depParser = depParser
self.requireEntities = requireEntities
self.parseName = parseName
ToolChain.__init__(self, steps)
self.modelParameterStringName = "preprocessorParams"
def __init__(self):
ToolChain.__init__(self)
# Steps
self.addStep("CONVERT", self.convert, {
"dataSetNames": None,
"corpusName": None
}, "documents.xml")
self.addStep("SPLIT-SENTENCES",
Tools.GeniaSentenceSplitter.makeSentences, {
"debug": False,
"postProcess": True
}, "sentences.xml")
self.addStep(
"NER", Tools.BANNER.run, {
"elementName": "entity",
"processElement": "sentence",
"debug": False,
"splitNewlines": True
}, "ner.xml")
self.addStep("PARSE", Tools.CharniakJohnsonParser.parse, {
"parseName": "McCC",
"requireEntities": False,
"debug": False
}, "parse.xml")
self.addStep("CONVERT-PARSE", Tools.StanfordParser.convertXML, {
"parser": "McCC",
"debug": False
}, "converted-parse.xml")
self.addStep("SPLIT-NAMES", ProteinNameSplitter.mainFunc,
{"parseName": "McCC"}, "split-names.xml")
self.addStep("FIND-HEADS", FindHeads.findHeads, {
"parse": "McCC",
"removeExisting": True
}, "heads.xml")
self.addStep("DIVIDE-SETS", self.divideSets, {
"outputStem": None,
"saveCombined": True
})
def __init__(self):
ToolChain.__init__(self)
self.modelParameterStringName = "preprocessorParams"
def __init__(self):
ToolChain.__init__(self)
self.modelParameterStringName = "preprocessorParams"
