def _ensureWordInDatabase(self, cursor, word):
db_words = Word.selectWhere(cursor, word=word)
if not db_words:
db_word = Word(word=word,
claim_count=0.,
download_count=0.)
db_word.commit(cursor, update=False)
def _ensureWordInDatabase(self, cursor, word):
db_words = Word.selectWhere(cursor, word=word)
if not db_words:
db_word = Word(word=word,
claim_count=0.,
download_count=0.)
db_word.commit(cursor, update=False)
def _updateDownloadStatistics(self, document, words):
cursor = self._cnx.cursor()
document.download_count = max(0, document.download_count) + 1
document.commit(cursor, update=True)
db_document_id = document.db_document_id
scores = {}
wordInfo = {}
for word in words:
scores[word] = DocumentScore.selectOrInsertWhere(
cursor, db_document_id=db_document_id, word=word)[0]
wordInfo[word] = Word.selectOrInsertWhere(cursor, word=word)[0]
for winfo in wordInfo.itervalues():
winfo.download_count += 1 / len(words)
winfo.commit(cursor, update=True)
for word, score in scores.iteritems():
score.download_count = max(0,
score.download_count) + 1 / len(words)
winfo_downloads = wordInfo[word].download_count
score.popularity = score.download_count / winfo_downloads
score.popularity -= hoeffding_deviation(winfo_downloads)
score.relevance = score.download_count / document.download_count
score.relevance -= hoeffding_deviation(document.download_count)
score.commit(cursor, update=True)
cursor.close()
self._cnx.commit()
def _updateDownloadStatistics(self, document, words):
cursor = self._cnx.cursor()
document.download_count = max(0, document.download_count) + 1
document.commit(cursor, update=True)
db_document_id = document.db_document_id
scores = {}
wordInfo = {}
for word in words:
scores[word] = DocumentScore.selectOrInsertWhere(cursor,
db_document_id=db_document_id,
word=word)[0]
wordInfo[word] = Word.selectOrInsertWhere(cursor,
word=word)[0]
for winfo in wordInfo.itervalues():
winfo.download_count += 1 / len(words)
winfo.commit(cursor, update=True)
for word,score in scores.iteritems():
score.download_count = max(0, score.download_count) + 1.0 / len(words)
winfo_downloads = wordInfo[word].download_count
score.popularity = float(score.download_count) / winfo_downloads
score.popularity -= hoeffding_deviation(winfo_downloads)
score.popularity = max(1e-6, score.popularity)
score.relevance = float(score.download_count) / document.download_count
score.relevance -= hoeffding_deviation(document.download_count)
score.relevance = max(1e-6, score.relevance)
score.commit(cursor, update=True)
cursor.close()
self._cnx.commit()
def _updateQueryStatistics(self, words):
# FIXME: update node_interests too
cursor = self._cnx.cursor()
for word in words:
winfo = Word.selectOrInsertWhere(cursor, word=word)[0]
winfo.claim_count += 1 / len(words)
winfo.commit(cursor, update=True)
cursor.close
self._cnx.commit()
def _updateQueryStatistics(self, words):
# FIXME: update node_interests too, but we need the nodeId to do this
cursor = self._cnx.cursor()
for word in words:
winfo = Word.selectOrInsertWhere(cursor, word=word)[0]
winfo.claim_count += 1 / len(words)
winfo.commit(cursor, update=True)
cursor.close()
self._cnx.commit()