Python 在同一台机器上使用相同输入时，NLTK输出会发生变化

下一段代码使用相同的inputself.synset返回不同的输出 为什么会这样？我做错什么了吗？还是由python引起的

def FilterSynSets(self):
    self.filteredSysNets = {}
    for synset in self.SysNets:
        for subsynset in self.SysNets:
            length = wn.path_similarity(synset,subsynset)
            if not length is None\
                and length !=1\
                and length >0:
                    target = synset.__str__().replace("'","")
                    source =subsynset.__str__().replace("'","")

                    connection="\"{}\"->\"{}\" [label={}]".format(
                        target,
                        source,
                        str(round(length,3)))

                    self.filteredSysNets[connection] = length
    oldLength = len(self.filteredSysNets)
    avarageVal = sum(self.filteredSysNets.values())/len(self.filteredSysNets)
    self.filteredSysNets = {k: v for k,v in self.filteredSysNets.items() if v>=avarageVal}
    newLength = len(self.filteredSysNets)
    prt = newLength/oldLength*100
    print ("avr -{}\n old -{}\n new - {}\n prec={}".format(avarageVal,oldLength,newLength,prt))
    return self

产出：

---

代码似乎是确定性的，因此我假设self.SysNets在运行之间会发生某种变化。Nope。self.Sysnets在input.BTW上总是相同的，它是synset而不是Sysnets=