lv1posSTR = lv1posSTR + re.sub(removePat1, "", i).replace("</", "(").replace(">", ") ")
print(lv1posSTR) #注意，lv11 會把 Verb 和 ASPECT (時態標記) 獨立切開，類比成英文就是 create-ed 視為 create / ed
lv2DICT = lv2DICT(inputSTR)
lv2posSTR = ""
for i in lv2DICT["result_pos"]:
if len(i)==1:
pass
else:
lv2posSTR = lv2posSTR + re.sub(removePat1, "", i).replace("</", "(").replace(">", ") ")
print(lv2posSTR) #注意，lv2 會把 Verb 和 ASPECT (時態標記) 結合起來，類比成英文就是 create+ed 成為 created
# 如果只是想取出動詞的「原型」的話，可以利用 .getVerbStemLIST() 的函式操作。
verbStemLIST = articut.getVerbStemLIST(lv1DICT)
print(verbStemLIST)
verbStemLIST = articut.getVerbStemLIST(lv2DICT)
print(verbStemLIST)
# 以上句法結合律說明結束
# 以下利用後處理，將「了」獨立切開
resultLIST = []
lv1posLIST = lv1posSTR.split(" ")
lv2posLIST = lv2posSTR.split(" ")
for i in range(0, len(lv2posLIST)):
if "了(VerbP)" in lv2posLIST[i]: #這裡是以「了」為例子。另一個中文的 -ed ASPECT 詞是「過」，列在下一項
resultLIST.extend(lv2posLIST[i].replace("了(VerbP)", "(ACTION_verb) 了(VerbP)").split(" "))
riseSignal = set(riseSignal)
#把 downSignal 和 riseSignal 中重覆的動詞清除。它可能是「中性」或是無關漲跌的動詞。
downSignal = downSignal - riseSignal.intersection(downSignal)
riseSignal = riseSignal - riseSignal.intersection(downSignal)
return (downSignal, riseSignal)
if __name__ == "__main__":
downSignal, riseSignal = signalMaker()
if None in (downSignal, riseSignal):
print("Cannot proceed!")
else:
testSTR = "產業供應鏈分散效應看好東協布局" #測試用句。注意到這一句並沒有在前述學習的 downSample/riseSample 中。
testResult = atc.parse(testSTR, level="lv2")
testVerbLIST = atc.getVerbStemLIST(testResult)
resultLIST = []
for tv in testVerbLIST:
if len(tv) == 0:
pass
else:
for v in tv:
if v[-1] in downSignal:
if "negation" in testResult["result_pos"][0][
v[0] - 22:
v[0]]: #確認是否有「否定詞」出現在 downSignal 中。如果有的話，那就是上漲囉！
resultLIST.append("這句新聞標題…應該是看漲↗")
else:
resultLIST.append("這句新聞標題…應該是看跌↘")
elif v[-1] in riseSignal:
if "negation" in testResult["result_pos"][0][
from ArticutAPI import Articut
text = "研究這個研究的研究已經被研究許多年了"
articut = Articut(username="", apikey="")
result = articut.parse(text)
#Articut 的 POS 標記斷詞結果在 "result_pos" 中。
print(result["result_pos"])
print("有幾個動詞的「研究」呢？")
verbLIST = articut.getVerbStemLIST(result, indexWithPOS=False)
verbCounter = 0
for sentence in verbLIST: #verbLIST 中，每「一個」句子是一個獨立的 list，故要先進入 sentence，再計算其中的 "研究"
for v in sentence:
if v[-1] == "研究":
print("發現動詞「研究」 ，位於原句的 {}~{} 位置".format(v[0], v[1]))
verbCounter = verbCounter + 1
print("共有 {} 個『研究』是動詞。".format(verbCounter))
print("有幾個名詞的「研究」呢？")
nounLIST = articut.getNounStemLIST(result, indexWithPOS=False)
def articut4PatentBot(categoryFILE, inputSTR):
with open("account.info", encoding="utf-8") as f:
userinfoDICT = json.loads(f.read())
articut = Articut(username=userinfoDICT["username"], apikey=userinfoDICT["apikey"], level="lv1")
# 讀入對應類別的專利文本
#patentDICT = patent[categoryFILE]
#patent_file = categoryFILE + '.json'
with open("Dataset\patent.json", encoding='utf-8') as f:
AllpatentDICT = json.loads(f.read())
patentDICT = AllpatentDICT[categoryFILE]
CertificateNumber = list(patentDICT.keys())
# 接收使用者輸入的專利範圍
userSTR = inputSTR.replace(" ", "").replace("\n", "")
userResultDICT = articut.parse(userSTR)
# 將類別中的專利範圍全部比對一次
VerbCosineSimilarity = []
NounCosineSimilarity = []
TFIDFCosineSimilarity = []
for k in patentDICT.values():
STR = k.replace(" ", "").replace("\n", "")
STRResultDICT = articut.parse(STR)
# 取得「動詞」做為特徵列表
patentVerbLIST = articut.getVerbStemLIST(STRResultDICT)
userVerbLIST = articut.getVerbStemLIST(userResultDICT)
# 利用 Counter() 模組計算每個動詞出現的次數
patentCOUNT = Counter(wordExtractor(patentVerbLIST, unify=True))
userCOUNT = Counter(wordExtractor(userVerbLIST, unify=True))
# 計算 [專利文本 vs. 使用者輸入文本] 的餘弦相似度
patent2userSIM = counterCosineSimilarity(patentCOUNT, userCOUNT)
VerbCosineSimilarity.append(patent2userSIM)
# 取得「名詞」做為特徵列表
patentNounLIST = articut.getNounStemLIST(STRResultDICT)
userNounLIST = articut.getNounStemLIST(userResultDICT)
# 利用 Counter() 模組計算每個名詞出現的次數
patentCOUNT = Counter(wordExtractor(patentNounLIST, unify=True))
userCOUNT = Counter(wordExtractor(userNounLIST, unify=True))
# 計算 [專利文本 vs. 使用者輸入文本] 的餘弦相似度
patent2userSIM = counterCosineSimilarity(patentCOUNT, userCOUNT)
NounCosineSimilarity.append(patent2userSIM)
# 取得「TF-IDF」做為特徵列表
patentTFIDFLIST = articut.analyse.extract_tags(STRResultDICT)
userTFIDFLIST = articut.analyse.extract_tags(userResultDICT)
# 利用 Counter() 模組計算每個 TF-IDF 特徵詞出現的次數
patentCOUNT = Counter(patentTFIDFLIST)
userCOUNT = Counter(userTFIDFLIST)
# 計算 [專利文本 vs. 使用者輸入文本] 的 TF-IDF 餘弦相似度
patent2userSIM = counterCosineSimilarity(patentCOUNT, userCOUNT)
TFIDFCosineSimilarity.append(patent2userSIM)
ArticutresultDICT = {}
max_Verb = max(VerbCosineSimilarity)
v = VerbCosineSimilarity.index(max_Verb)
# print("[專利文本 vs. 使用者輸入文本] 的動詞餘弦相似度:{}".format(VerbCosineSimilarity))
# print("最大值為{:.2f}來自證書書號{}的專利範圍".format(max_Verb, CertificateNumber[v]))
ArticutresultDICT["Verb"] = {}
ArticutresultDICT["Verb"][CertificateNumber[v]] = max_Verb
max_Noun = max(NounCosineSimilarity)
n = NounCosineSimilarity.index(max_Noun)
# print("[專利文本 vs. 使用者輸入文本] 的名詞餘弦相似度:{}".format(NounCosineSimilarity))
# print("最大值為{:.2f}來自證書書號{}的專利範圍".format(max_Noun, CertificateNumber[n]))
ArticutresultDICT["Noun"] = {}
ArticutresultDICT["Noun"][CertificateNumber[n]] = max_Noun
max_TFIDF = max(TFIDFCosineSimilarity)
t = TFIDFCosineSimilarity.index(max_TFIDF)
# print("[專利文本 vs. 使用者輸入文本] 的 TF-IDF 特徵詞餘弦相似度:{}".format(TFIDFCosineSimilarity))
# print("最大值為{:.2f}來自證書書號{}的專利範圍".format(max_TFIDF, CertificateNumber[t]))
ArticutresultDICT["TFIDF"] = {}
ArticutresultDICT["TFIDF"][CertificateNumber[t]] = max_TFIDF
ArticutresultDICT["All_Max"] = {}
m = max(max_Verb, max_Noun, max_TFIDF)
if m == max_Noun:
ArticutresultDICT["All_Max"][CertificateNumber[n]] = [m, "名詞"]
elif m == max_Verb:
ArticutresultDICT["All_Max"][CertificateNumber[v]] = [m, "動詞"]
elif m == max_TFIDF:
ArticutresultDICT["All_Max"][CertificateNumber[t]] = [m, "TF-IDF"]
return ArticutresultDICT
if __name__ == "__main__":
articut = Articut(username="******",
apikey="4#-!FX^Vr5kBaznKL2U7egRwkU=Hx*k")
fileTUPLE = ("../example/text.txt", "./A. forsteri.txt")
MouseSTR = jsonTextReader(fileTUPLE[0])
#print(MouseSTR)
PenguinSTR = jsonTextReader(fileTUPLE[1])
#print(PenguinSTR)
MouseDICT_lv3 = EventAnalysis(MouseSTR, articut)
MouseLIST = MouseDICT_lv3["event"]
MouseDICT_lv2 = articut.parse(MouseSTR, level="lv2")
MouseLIST_lv2 = articut.getVerbStemLIST(MouseDICT_lv2)
#for item in MouseLIST_lv2:
# if item != '\n' and item != []:
# print(item)
PenguinDICT_lv3 = EventAnalysis(PenguinSTR, articut)
PenguinLIST = PenguinDICT_lv3["event"]
ResultDICT = {"倉鼠": [], "皇帝企鵝": []}
for item in MouseLIST:
if item != '\n' and item != []:
ResultDICT["倉鼠"].append(item)
for item in PenguinLIST:
if item != '\n' and item != []:
ResultDICT["皇帝企鵝"].append(item)
print(ResultDICT)