Allow search by name as well as url

poloclub · Jun 13, 2021 · 3b9bed8 · 3b9bed8
1 parent a31c719
commit 3b9bed8
Show file tree

Hide file tree

Showing 15 changed files with 11,002 additions and 38 deletions.
diff --git a/Preprocessing/NewGoogleQueryCited.py b/Preprocessing/NewGoogleQueryCited.py
@@ -37,7 +37,7 @@ def newGenerateCitedGoogleScholarCSV(list_of_researchers):
 
     	#researcher = list_of_researchers.iloc[i][0]
     	researcher = ""
-    	researcher_URL = list_of_researchers.iloc[i][0]
+    	researcher_URL = list_of_researchers.iloc[i][1]
 
     	researcherID = researcher_URL.replace("https://scholar.google.com/citations?user=", "")
     	researcherID = researcherID.replace("&hl=en&oi=ao", "")
@@ -72,10 +72,38 @@ def newGenerateCitedGoogleScholarCSV(list_of_researchers):
 	    		outputGrid.append([researcher, str(researcher_URL), "False"])
 
     	except:
-    		search_query = 'error'
-    		author = 'error'
 
-    		outputGrid.append([researcher, str(researcher_URL), "False"])
+    		try:
+
+    			search_query = next(scholarly.search_author(list_of_researchers.iloc[i][0]))
+	    		author = search_query.fill()
+
+
+
+	    		researcher = author.name
+
+	    		print(researcher)
+
+
+
+	    		if author != None:
+
+	    			publications = author.publications
+
+	    			outputGrid.append([researcher, str(researcher_URL), "True"])
+
+	    		else:
+	    			search_query = 'error'
+		    		author = 'error'
+
+		    		outputGrid.append([researcher, str(researcher_URL), "False"])
+
+    		except:
+
+	    		search_query = 'error'
+	    		author = 'error'
+
+	    		outputGrid.append([researcher, str(researcher_URL), "False"])
 
 
     	if search_query != 'error':

diff --git a/Preprocessing/NewGoogleQueryRecent.py b/Preprocessing/NewGoogleQueryRecent.py
@@ -180,7 +180,7 @@ def newGenerateRecentGoogleScholarCSV(list_of_researchers):
 
 		#researcher = list_of_researchers.iloc[i][0]
 		researcher = ""
-		researcher_URL = list_of_researchers.iloc[i][0]
+		researcher_URL = list_of_researchers.iloc[i][1]
 
 		researcherID = researcher_URL.replace("https://scholar.google.com/citations?user=", "")
 		researcherID = researcherID.replace("&hl=en&oi=ao", "")
@@ -214,10 +214,37 @@ def newGenerateRecentGoogleScholarCSV(list_of_researchers):
 
 		except:
 
-			search_query = 'error'
-			author = 'error'
+			try:
 
-			outputGrid.append([researcher, str(researcher_URL), "False"])
+	    		search_query = next(scholarly.search_author(list_of_researchers.iloc[i][0]))
+	    		author = search_query.fill()
+
+
+
+	    		researcher = author.name
+
+	    		print(researcher)
+
+
+
+	    		if author != None:
+
+	    			publications = author.publications
+
+	    			outputGrid.append([researcher, str(researcher_URL), "True"])
+
+	    		else:
+	    			search_query = 'error'
+		    		author = 'error'
+
+		    		outputGrid.append([researcher, str(researcher_URL), "False"])
+
+    		except:
+
+	    		search_query = 'error'
+	    		author = 'error'
+
+	    		outputGrid.append([researcher, str(researcher_URL), "False"])
 
 
 		if search_query != 'error':

diff --git a/ResearchersDataset.csv b/ResearchersDataset.csv
@@ -1,11 +1,21 @@
-URL
-https://scholar.google.com/citations?user=B3U7yvcAAAAJ&hl=it
-https://scholar.google.com/citations?user=JnkeH28AAAAJ&hl=it
-https://scholar.google.co.uk/citations?user=sbcJOSoAAAAJ&hl=en
-https://scholar.google.co.uk/citations?user=JXTpQXIAAAAJ&hl=en
-https://scholar.google.com.mx/citations?user=SPrHADoAAAAJ&hl=en
-https://scholar.google.co.uk/citations?user=o2gCskMAAAAJ&hl=en
-https://scholar.google.ch/citations?user=IoKBsjcAAAAJ&hl=en
-https://scholar.google.com/citations?user=uE1oO1gAAAAJ&hl=en
-https://scholar.google.com/citations?user=vNvRYksAAAAJ&hl=en
-https://scholar.google.com/citations?user=wIFFAosAAAAJ&hl=en
+Name,URL
+Guido Noto La Diega,https://scholar.google.com/citations?user=B3U7yvcAAAAJ&hl=it
+Rossana Ducato,https://scholar.google.com/citations?user=JnkeH28AAAAJ&hl=it
+Martin Kretschmer,https://scholar.google.co.uk/citations?user=sbcJOSoAAAAJ&hl=en
+Marta Iljadica,https://scholar.google.co.uk/citations?user=JXTpQXIAAAAJ&hl=en
+Israel Cedillo Lazcano,https://scholar.google.com.mx/citations?user=SPrHADoAAAAJ&hl=en
+Dr Lachlan Urquhart,https://scholar.google.co.uk/citations?user=o2gCskMAAAAJ&hl=en
+Mariela de Amstalden,https://scholar.google.ch/citations?user=IoKBsjcAAAAJ&hl=en
+Jade Kouletakis,https://scholar.google.com/citations?user=uE1oO1gAAAAJ&hl=en
+"Irene Couzigou, PhD, LLM",https://scholar.google.com/citations?user=vNvRYksAAAAJ&hl=en
+Patricia Živković,https://scholar.google.com/citations?user=wIFFAosAAAAJ&hl=en
+Emma Milne,https://scholar.google.co.uk/citations?user=iStd81MAAAAJ&hl=en
+C William R Webster,https://scholar.google.co.uk/citations?user=mdypkb4AAAAJ&hl=en
+Péter Cserne,https://scholar.google.com/citations?user=qlvq1lwAAAAJ&hl=en
+Dr Mo Egan,https://scholar.google.co.uk/citations?user=6o0j57EAAAAJ&hl=en
+Lilian Edwards,https://scholar.google.com/citations?user=a6-onLoAAAAJ&hl=en
+Matthew Jewell,https://scholar.google.co.uk/citations?user=OkUv-F4AAAAJ&hl=en
+Laurence Diver,https://scholar.google.co.uk/citations?user=9L-xODcAAAAJ&hl=en
+Gianluca Andresani,https://scholar.google.com/citations?user=poIiHR0AAAAJ&hl=en
+Burkhard Schafer,https://scholar.google.com/citations?user=vD2DRqoAAAAJ&hl=en
+Paolo Cavaliere,https://scholar.google.com/citations?user=HOYJ86sAAAAJ&hl=en
diff --git a/ResearchersScrapedCited.csv b/ResearchersScrapedCited.csv
@@ -1,10 +1,20 @@
 Guido Noto La Diega,https://scholar.google.com/citations?user=B3U7yvcAAAAJ&hl=it,True
 Rossana Ducato,https://scholar.google.com/citations?user=JnkeH28AAAAJ&hl=it,True
-,https://scholar.google.co.uk/citations?user=sbcJOSoAAAAJ&hl=en,False
-,https://scholar.google.co.uk/citations?user=JXTpQXIAAAAJ&hl=en,False
-,https://scholar.google.com.mx/citations?user=SPrHADoAAAAJ&hl=en,False
-,https://scholar.google.co.uk/citations?user=o2gCskMAAAAJ&hl=en,False
-,https://scholar.google.ch/citations?user=IoKBsjcAAAAJ&hl=en,False
+Martin Kretschmer,https://scholar.google.co.uk/citations?user=sbcJOSoAAAAJ&hl=en,True
+Marta Iljadica,https://scholar.google.co.uk/citations?user=JXTpQXIAAAAJ&hl=en,True
+Israel Cedillo Lazcano,https://scholar.google.com.mx/citations?user=SPrHADoAAAAJ&hl=en,True
+Dr Lachlan Urquhart,https://scholar.google.co.uk/citations?user=o2gCskMAAAAJ&hl=en,True
+Mariela de Amstalden,https://scholar.google.ch/citations?user=IoKBsjcAAAAJ&hl=en,True
 Jade Kouletakis,https://scholar.google.com/citations?user=uE1oO1gAAAAJ&hl=en,True
 Irene Couzigou, PhD, LLM,https://scholar.google.com/citations?user=vNvRYksAAAAJ&hl=en,True
 Patricia Živković,https://scholar.google.com/citations?user=wIFFAosAAAAJ&hl=en,True
+Dafouz Milne, Emma,https://scholar.google.co.uk/citations?user=iStd81MAAAAJ&hl=en,True
+C William R Webster,https://scholar.google.co.uk/citations?user=mdypkb4AAAAJ&hl=en,True
+Péter Cserne,https://scholar.google.com/citations?user=qlvq1lwAAAAJ&hl=en,True
+Dr Mo Egan,https://scholar.google.co.uk/citations?user=6o0j57EAAAAJ&hl=en,True
+Lilian Edwards,https://scholar.google.com/citations?user=a6-onLoAAAAJ&hl=en,True
+Matthew Jewell,https://scholar.google.co.uk/citations?user=OkUv-F4AAAAJ&hl=en,True
+Dominik Ziegler,https://scholar.google.co.uk/citations?user=9L-xODcAAAAJ&hl=en,True
+Gianluca Andresani,https://scholar.google.com/citations?user=poIiHR0AAAAJ&hl=en,True
+Burkhard Schafer,https://scholar.google.com/citations?user=vD2DRqoAAAAJ&hl=en,True
+Paolo Cavaliere,https://scholar.google.com/citations?user=HOYJ86sAAAAJ&hl=en,True
diff --git a/ResearchersScrapedRecent.csv b/ResearchersScrapedRecent.csv
@@ -1,10 +1,20 @@
 Guido Noto La Diega,https://scholar.google.com/citations?user=B3U7yvcAAAAJ&hl=it,True
 Rossana Ducato,https://scholar.google.com/citations?user=JnkeH28AAAAJ&hl=it,True
-,https://scholar.google.co.uk/citations?user=sbcJOSoAAAAJ&hl=en,False
-,https://scholar.google.co.uk/citations?user=JXTpQXIAAAAJ&hl=en,False
-,https://scholar.google.com.mx/citations?user=SPrHADoAAAAJ&hl=en,False
-,https://scholar.google.co.uk/citations?user=o2gCskMAAAAJ&hl=en,False
-,https://scholar.google.ch/citations?user=IoKBsjcAAAAJ&hl=en,False
+Martin Kretschmer,https://scholar.google.co.uk/citations?user=sbcJOSoAAAAJ&hl=en,True
+Marta Iljadica,https://scholar.google.co.uk/citations?user=JXTpQXIAAAAJ&hl=en,True
+Israel Cedillo Lazcano,https://scholar.google.com.mx/citations?user=SPrHADoAAAAJ&hl=en,True
+Dr Lachlan Urquhart,https://scholar.google.co.uk/citations?user=o2gCskMAAAAJ&hl=en,True
+Mariela de Amstalden,https://scholar.google.ch/citations?user=IoKBsjcAAAAJ&hl=en,True
 Jade Kouletakis,https://scholar.google.com/citations?user=uE1oO1gAAAAJ&hl=en,True
 Irene Couzigou, PhD, LLM,https://scholar.google.com/citations?user=vNvRYksAAAAJ&hl=en,True
 Patricia Živković,https://scholar.google.com/citations?user=wIFFAosAAAAJ&hl=en,True
+Dafouz Milne, Emma,https://scholar.google.co.uk/citations?user=iStd81MAAAAJ&hl=en,True
+C William R Webster,https://scholar.google.co.uk/citations?user=mdypkb4AAAAJ&hl=en,True
+Péter Cserne,https://scholar.google.com/citations?user=qlvq1lwAAAAJ&hl=en,True
+Dr Mo Egan,https://scholar.google.co.uk/citations?user=6o0j57EAAAAJ&hl=en,True
+Lilian Edwards,https://scholar.google.com/citations?user=a6-onLoAAAAJ&hl=en,True
+Matthew Jewell,https://scholar.google.co.uk/citations?user=OkUv-F4AAAAJ&hl=en,True
+Dominik Ziegler,https://scholar.google.co.uk/citations?user=9L-xODcAAAAJ&hl=en,True
+Gianluca Andresani,https://scholar.google.com/citations?user=poIiHR0AAAAJ&hl=en,True
+Burkhard Schafer,https://scholar.google.com/citations?user=vD2DRqoAAAAJ&hl=en,True
+Paolo Cavaliere,https://scholar.google.com/citations?user=HOYJ86sAAAAJ&hl=en,True
diff --git a/citedScholarDataset.gzip b/citedScholarDataset.gzip
diff --git a/package-lock.json b/package-lock.json
diff --git a/recentScholarDataset.gzip b/recentScholarDataset.gzip