def simTextRank(filePath, summarySentenceCount):
'''
summary generation using similarity between sentences
'''
global sentence_dictionary
sentence_dictionary = {}
sentences = []
sentence_dictionary, sentences, size = cleanText(
filePath) #input after preprocessing
graph = generateGraph(list(
sentence_dictionary.keys())) #keys are sentence ids
pageRank = networkx.pagerank(
graph) #computes ranking of nodes in graph,return type is a dictionary
output = "\n".join([
sentences[sentenceID] for sentenceID in sorted(
sorted(pageRank, key=pageRank.get, reverse=True)
[:summarySentenceCount])
])
with open(os.path.join(app.config['DOWNLOAD_FOLDER'], 'sim_textrank.txt'),
"w",
encoding="utf-8") as outFile:
outFile.write(output)
outFile.close()
def process(arg1, arg2, arg3):
'''
:param arg1: path to the file containing the text to be summarized
:param arg2: Number of sentences to be extracted as summary
:param arg3: size of the window to be used in the co-occurance relation
'''
global window, n, numberofSentences, textRank, sentenceDictionary, size, sentences
if arg1 != None and arg2 != None and arg3 != None:
sentenceDictionary, sentences, size = cleanText(arg1)
window = int(arg3)
numberofSentences = int(arg2)
n = int(math.ceil(min(0.1 * size, 7 * math.log(size))))
generatepositionaldistribution()
keyphrases = textrank()
summarize(arg1, keyphrases, numberofSentences)
else:
print("not enough parameters")
def textRankSimilarity(filePath, summarySentenceCount):
global sentenceDictionary
sentenceDictionary = {}
sentences = []
sentenceDictionary, sentences, size = cleanText(filePath)
#printDictionary()
graph = generateGraph(list(sentenceDictionary.keys()))
pageRank = networkx.pagerank(graph)
output = "\n".join([
sentences[sentenceID] for sentenceID in sorted(
sorted(pageRank, key=pageRank.get, reverse=True)
[:summarySentenceCount])
])
print("\nSummary:")
print(output)
with io.open("../Marathi/summaries/" + (filePath).split('/')[-1] +
"_TextRankSimilaritySummarizer",
"w",
encoding='utf-8') as outFile:
outFile.write(output)
outFile.close()