def explainRelevance(guid):
"""
Given a guid, it prepares its explainer document
"""
doc=cp.Corpus.loadSciDoc(guid)
html=doc.prettyPrintDocumentHTML(True,True,False,tokenizeAndTag, referenceFormatting)
computeOverlapForAllReferences(doc,doc.glob["_inverted_index"])
writeFileText(html,cp.Corpus.dir_output+guid+"_explain.html")
def explainRelevance(guid):
"""
Given a guid, it prepares its explainer document
"""
doc = cp.Corpus.loadSciDoc(guid)
html = doc.prettyPrintDocumentHTML(True, True, False, tokenizeAndTag,
referenceFormatting)
computeOverlapForAllReferences(doc, doc.glob["_inverted_index"])
writeFileText(html, cp.Corpus.dir_output + guid + "_explain.html")
def generateVisualization(self, guid):
"""
Given a guid, it prepares its explainer document
"""
## output_dir=Corpus.dir_output
output_dir=r"C:\Users\dd\Documents\Dropbox\PhD\code\doc_visualization\\"
doc=Corpus.loadSciDoc(guid)
Corpus.tagAllReferencesAsInCollectionOrNot(doc)
counts1=self.getDocumentTokens(doc)
# generate a unique id for each unique term, make a dictionary
for index, token in enumerate(counts1):
self.term_info[token]={"token_id":str(index), "references": []}
self.overlapping_tokens={}
in_collection_references=Corpus.getMetadataByGUID(guid)["outlinks"]
for ref in doc["references"]:
match=Corpus.matchReferenceInIndex(ref)
if match:
doc2=Corpus.loadSciDoc(match["guid"])
counts2=self.getDocumentTokens(doc2)
# for each in_collection_reference number (0 onwards) we store the list
# of its overlapping tokens with the current document
self.overlapping_tokens[ref["id"]]=self.getOverlappingTokens(counts1, counts2)
for token in self.overlapping_tokens[ref["id"]]:
ref_list=self.term_info[token]["references"]
if ref["id"] not in ref_list:
ref_list.append(ref["id"])
# try to find some signal in the noise
self.filterTokens()
json_str="var token_data=" + json.dumps(self.overlapping_tokens) + ";"
writeFileText(json_str, output_dir+guid+"_data.json")
html=doc.prettyPrintDocumentHTML(
True,
True,
False,
## extra_attribute_function=self.extraAttributes,
citation_formatting_function=self.citationFormatting,
reference_formatting_function=self.referenceFormatting,
text_formatting_function=self.textFormatting
)
html=self.padWithHTML(html, guid)
writeFileText(html,output_dir+guid+"_vis.html")
def generateVisualization(self, guid):
"""
Given a guid, it prepares its explainer document
"""
## output_dir=Corpus.dir_output
output_dir=r"C:\Users\dd\Documents\Dropbox\PhD\code\doc_visualization\\"
doc=Corpus.loadSciDoc(guid)
Corpus.tagAllReferencesAsInCollectionOrNot(doc)
counts1=self.getDocumentTokens(doc)
# generate a unique id for each unique term, make a dictionary
for index, token in enumerate(counts1):
self.term_info[token]={"token_id":str(index), "references": []}
self.overlapping_tokens={}
in_collection_references=Corpus.getMetadataByGUID(guid)["outlinks"]
for ref in doc["references"]:
match=Corpus.matchReferenceInIndex(ref)
if match:
doc2=Corpus.loadSciDoc(match["guid"])
counts2=self.getDocumentTokens(doc2)
# for each in_collection_reference number (0 onwards) we store the list
# of its overlapping tokens with the current document
self.overlapping_tokens[ref["id"]]=self.getOverlappingTokens(counts1, counts2)
for token in self.overlapping_tokens[ref["id"]]:
ref_list=self.term_info[token]["references"]
if ref["id"] not in ref_list:
ref_list.append(ref["id"])
# try to find some signal in the noise
self.filterTokens()
json_str="var token_data=" + json.dumps(self.overlapping_tokens) + ";"
writeFileText(json_str, output_dir+guid+"_data.json")
html=doc.prettyPrintDocumentHTML(
True,
True,
False,
## extra_attribute_function=self.extraAttributes,
citation_formatting_function=self.citationFormatting,
reference_formatting_function=self.referenceFormatting,
text_formatting_function=self.textFormatting
)
html=self.padWithHTML(html, guid)
writeFileText(html,output_dir+guid+"_vis.html")
