def generateWeeklyWC (cname, prefix, smartStopWords):
p = PostGetter()
minWeek = p.getMinWeekForCourse(cname)
maxWeek = p.getMaxWeekForCourse(cname)
for week in range(minWeek, maxWeek+1):
numPosts = p.getNumPosts(cname, week)
if (numPosts < 20):
continue
text=p.getWeeklyDataForCourseAsString(cname, week)
num = (week-minWeek)
(wordcountDict, w_, c_)=getWordCounts(text,'%s_week%s_stop'%(prefix,week),'%s_week%s'%(prefix, week), smartStopWords)
generateWC(w_,c_,'%s_week%s_stop'%(prefix,week))
def generatePenalizeWeeklyWindows (cname, prefix, diff):
p = PostGetter ()
minWeek = p.getMinWeekForCourse (cname)
maxWeek = p.getMaxWeekForCourse (cname)
d = {}
for week in range(minWeek, maxWeek):
currText = p.getWeeklyDataForCourseAsString(cname, week)
smartStopWords = True
(currWordCountDict, currw_, currc_) = getWordCounts(currText,'%s_week%s_stop'%(prefix,week),'%s_week%s'%(prefix, week), smartStopWords)
d[week] = currWordCountDict
for week in range(minWeek+1, maxWeek):
currWordCountDict = d[week]
if week == 201412:
print "CURRRRRRRRRRRRR"
#raw_input('bar')
print currWordCountDict
#raw_input ('foo')
ctr=diff
for i in range (week-1, minWeek-1,-1) :
ctr= ctr-1
if (ctr == 0):
break
prevDict = d[i]
diffDict = weightWords (prevDict, currWordCountDict, 0.5)
currWordCountDict = diffDict
print currWordCountDict
if week == 201412:
#raw_input ('aa')
print currWordCountDict
#raw_input ('aa')
if week == 201412:
print "CURRRRRRRRRRRRR"
#raw_input('end')
print currWordCountDict
#raw_input ('end')
(w_,c_)=dictToArray (diffDict)
print w_
print c_
print diff
if (len(w_) > 3):
generateWC(w_,c_,'%s_week%s_stop_DIFF_EXP_PENALTY_DIFF%s'%(prefix,week,diff))
def main():
p=PostGetter()
#cname='Engineering/CVX101/Winter2014'
#cname='Education/EDUC115N/How_to_Learn_Math'
#cname = 'Medicine/HRP258/Statistics_in_Medicine'
if len(sys.argv) < 2:
print 'Usage: python wordcloud.py <cname> sample course names: Engineering/CVX101/Winter2014,Medicine/HRP258/Statistics_in_Medicine'
sys.exit(1)
cname = sys.argv[1]
text = p.getAllDataForCourseAsString(cname)
print text
smartStopWords = True
#(wordcountDict,w_,c_)=getWordCounts(text,'wc_cvx_full_stop','wc_cvx_full', smartStopWords)
#generateWC(w_,c_,'wc_cvx_full_stop')
#generateWeeklyWC(cname, 'WEEKLYwc_cvx', smartStopWords)
generateDiff (cname,'WEEKL_DIFF_lmath', smartStopWords)
def generateDiff (cname, prefix, smartStopWords):
p = PostGetter()
minWeek = p.getMinWeekForCourse(cname)
maxWeek = p.getMaxWeekForCourse(cname)
for week in range(minWeek, maxWeek):
curr = week
next = week+1
currNumPosts = p.getNumPosts(cname, curr)
nextNumPosts = p.getNumPosts(cname, next)
if (currNumPosts < 20 or nextNumPosts < 20):
continue
currText = p.getWeeklyDataForCourseAsString(cname, curr)
nextText = p.getWeeklyDataForCourseAsString(cname, next)
(currWordCountDict, currw_, currc_) = getWordCounts(currText,'%s_week%s_stop'%(prefix,curr),'%s_week%s'%(prefix, curr), smartStopWords)
(nextWordCountDict, nextw_, nextc_) = getWordCounts(nextText,'%s_week%s_stop'%(prefix,next),'%s_week%s'%(prefix, next), smartStopWords)
diffDict = getDiff (currWordCountDict, nextWordCountDict)
(w_,c_)=dictToArray (diffDict)
generateWC(w_,c_,'%s_week%s_stop_DIFF'%(prefix,week+1))
def generatePenalizeWeeks (cname, prefix, diff):
p = PostGetter ()
minWeek = p.getMinWeekForCourse (cname)
maxWeek = p.getMaxWeekForCourse (cname)
for week in range(minWeek, maxWeek):
curr = week
next = week+1
currNumPosts = p.getNumPosts(cname, curr)
nextNumPosts = p.getNumPosts(cname, next)
if (currNumPosts < 20 or nextNumPosts < 20):
continue
currText = p.getWeeklyDataForCourseAsString(cname, curr)
nextText = p.getWeeklyDataForCourseAsString(cname, next)
smartStopWords = True
(currWordCountDict, currw_, currc_) = getWordCounts(currText,'%s_week%s_stop'%(prefix,curr),'%s_week%s'%(prefix, curr), smartStopWords)
(nextWordCountDict, nextw_, nextc_) = getWordCounts(nextText,'%s_week%s_stop'%(prefix,next),'%s_week%s'%(prefix, next), smartStopWords)
penalty = 0.5
diffDict = weightWords (currWordCountDict, nextWordCountDict , penalty)
(w_,c_)=dictToArray (diffDict)
generateWC(w_,c_,'%s_week%s_stop_DIFFPENALTY%s'%(prefix,week+1,diff))
def generateDiffConfigurableWeeks (cname, prefix, diff):
p = PostGetter ()
minWeek = p.getMinWeekForCourse (cname)
maxWeek = p.getMaxWeekForCourse (cname)
for week in range(minWeek, maxWeek):
curr = week
next = week+diff
if next > maxWeek:
break
currNumPosts = p.getNumPosts(cname, curr)
nextNumPosts = p.getNumPosts(cname, next)
if (currNumPosts < 20 or nextNumPosts < 20):
continue
nextText = p.getWeeklyDataForCourseAsString (cname, next)
smartStopWords = True
(nextWordCountDict, nextw_, nextc_) = getWordCounts(nextText,'%s_week%s_stop'%(prefix,next),'%s_week%s'%(prefix, next), smartStopWords)
for i in range (curr, next):
currText = p.getWeeklyDataForCourseAsString(cname, i)
(currWordCountDict, currw_, currc_) = getWordCounts(currText,'%s_week%s_stop'%(prefix,i),'%s_week%s'%(prefix, i), smartStopWords)
diffDict = getDiff (currWordCountDict, nextWordCountDict)
nextWordCountDict = diffDict
(w_,c_)=dictToArray (diffDict)
generateWC(w_,c_,'%s_week%s_stop_DIFFCONF%s'%(prefix,week+1,diff))