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))
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 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))
