def getLevelWithConfNum():
mRedis = RedisHelper()
authors = mRedis.getAllAuthors()
confPaperNumDict = dict()
confCoauNumDict = dict()
index = 0
for author in authors:
index += 1
if index % 1000 == 0:
logging.info(index)
confNum = len(mRedis.getAuConfs(author))
paperNum = len(mRedis.getAuPapers(author))
coauNum = len(mRedis.getAuCoauthors(author))
paperNumList = confPaperNumDict.setdefault(confNum, [])
paperNumList.append(paperNum)
coauNumList = confCoauNumDict.setdefault(confNum, [])
coauNumList.append(coauNum)
index = 0
with open(OUTPUT_ACADEMIC_LEVEL_PAPER_COAU_NUM, 'w') as fileWriter:
for confNum in confPaperNumDict.keys():
index += 1
if index % 1000 == 0:
logging.info(index)
paperNumList = confPaperNumDict[confNum]
avg_P, max_num_P, min_num_P = statisticOfList(paperNumList)
coauNumList = confCoauNumDict[confNum]
avg_C, max_num_C, min_num_C = statisticOfList(coauNumList)
fileWriter.write(
str(confNum) + '\t' + str(avg_P) + '\t' + str(max_num_P) +
'\t' + str(min_num_P) + '\t' + str(avg_C) + '\t' +
str(max_num_C) + '\t' + str(min_num_C) + '\n')
fileWriter.close()
def __init__(self):
self.mRedis = RedisHelper()
self.isPaperTag = False
self.conf = ''
self.isTitleTag = False
self.paperId = -1
self.isAuthorTag = False
self.authors = list()
self.isYearTag = False
self.year = ''
def __init__(self):
self.G = nx.Graph()
self.stars = dict()
self.targets = dict()
self.loadStarsAndTargets()
logging.info('loadStarsAndTargets done---------------')
self.shortestPathLength = dict()
self.mRedis = RedisHelper()
self.buildGraph()
logging.info('---------------')
def getConfPaperNum():
"""统计每个会议收录了多少论文。
格式：会议名-->收录论文数
"""
index = 0
with open(OUTPUT_STATISTIC_CONF_PAPER_NUM, 'w') as fileWriter:
mRedis = RedisHelper()
confPaperNum = dict()
conferences = mRedis.getAllConfs()
for conf in conferences:
index += 1
logging.info(str(index) + '--' + conf)
paperNum = len(mRedis.getConfPapers(conf))
tmp = confPaperNum.setdefault(paperNum, 0)
confPaperNum[paperNum] = tmp + 1
for paperNum, confNum in confPaperNum.items():
fileWriter.write(str(paperNum) + '\t' + str(confNum) + '\n')
fileWriter.close()
def extracStarsAndTargets():
mRedis = RedisHelper()
starts = dict()
targets = dict()
authors = mRedis.getAllAuthors()
confNumAuthors = dict()
index = 0
for author in authors:
index += 1
if index % 1000 == 0:
logging.info(index)
confNum = len(mRedis.getAuConfs(author))
tmp = confNumAuthors.setdefault(confNum, [])
tmp.append(author)
logging.info('-------')
for i in range(1, 101):
logging.info(i)
confauhtors = confNumAuthors[i]
count = 0
while count < 10:
au = random.choice(confauhtors)
if au not in targets.keys():
targets[au] = i
count += 1
count = 0
while count < 100:
author = random.choice(authors)
confNum = len(mRedis.getAuConfs(author))
if author not in stars and confNum > 100:
stars[author] = confNum
count += 1
logging.info(count)
authors = []
confNumAuthors = {}
with open(OUTPUT_STAR_AUTHORS, 'w') as fileWriter:
for author, confNum in stars.items():
fileWriter.write(author + '\t' + str(confNum) + '\n')
fileWriter.close()
with open(OUTPUT_TARGET_AUTHORS, 'w') as fileWriter:
for author, confNum in targets.items():
fileWriter.write(author + '\t' + str(confNum) + '\n')
fileWriter.close()
def getAuthorConfNum():
"""统计每个学者的参会数。
格式：参会数-->学者数
"""
index = 0
with open(OUTPUT_STATISTIC_AUTHOR_CONF_NUM, 'w') as fileWriter:
mRedis = RedisHelper()
authors = mRedis.getAllAuthors()
authorConfNum = dict()
for author in authors:
index += 1
if index % 1000 == 0:
logging.info(index)
confNum = len(mRedis.getAuConfs(author))
tmp = authorConfNum.setdefault(confNum, 0)
authorConfNum[confNum] = tmp + 1
for k, v in authorConfNum.items():
fileWriter.write(str(k) + '\t' + str(v) + '\n')
fileWriter.close()
def getAuthorPaperNum():
"""统计每个学者发表了多少论文。
格式：论文数-->学者数
"""
index = 0
with open(OUTPUT_STATISTIC_AUTHOR_PAPER_NUM, 'w') as fileWriter:
mRedis = RedisHelper()
authors = mRedis.getAllAuthors()
authorPaperNum = dict()
for author in authors:
index += 1
if index % 1000 == 0:
logging.info(index)
paperNum = len(mRedis.getAuPapers(author))
tmp = authorPaperNum.setdefault(paperNum, 0)
authorPaperNum[paperNum] = tmp + 1
for k, v in authorPaperNum.items():
fileWriter.write(str(k) + '\t' + str(v) + '\n')
fileWriter.close()
def __init__(self):
self.mRedis = RedisHelper()
