overall coverage - 42%

1 files changed, 16 insertions, 4 deletions

diff --git a/s2-citation-report.py b/s2-citation-report.py
index 26e148fe..79e34a0e 100644
--- a/s2-citation-report.py
+++ b/s2-citation-report.py
@@ -11,13 +11,23 @@ from util import *
 def s2_citation_report():
   addresses = AddressBook()
   megapixels = load_megapixels_queries()
+  successful_geocodes = {}
   papers = []
   for fn in glob.iglob('datasets/s2/papers/**/*.json', recursive=True):
-    paper_data = process_paper(fn, addresses, megapixels)
+    paper_data = process_paper(fn, addresses, megapixels, successful_geocodes)
     papers.append(paper_data)
-  write_papers_report('reports/report_index.html', 'All Papers', papers, 'paperId')
+  write_papers_report('reports/report_index.html', 'All Papers', papers, 'title')
   write_papers_report('reports/report_coverage.html', 'Coverage', papers, 'citations_geocoded', reverse=True)
 
+  paper_count = 0
+  geocode_count = 0
+  for key, value in successful_geocodes.items():
+    if value:
+      geocode_count += 1
+    paper_count += 1
+  print("citations: {}".format(paper_count))
+  print("geocoded: {} ({}%)".format(geocode_count, percent(geocode_count, paper_count)))
+
 def write_papers_report(fn, title, papers, key, reverse=False):
   sorted_papers = []
   for paper in sorted(papers, key=lambda x: x[key], reverse=reverse):
@@ -31,7 +41,7 @@ def write_papers_report(fn, title, papers, key, reverse=False):
       paper['address'],
       paper['lat'],
       paper['lng'],
-      str(percent(paper['citation_count'], paper['citations_geocoded'])) + '%',
+      str(percent(paper['citations_geocoded'], paper['citation_count'])) + '%',
       paper['citation_count'],
       paper['citations_geocoded'],
       paper['citations_unknown'],
@@ -59,7 +69,7 @@ def write_papers_report(fn, title, papers, key, reverse=False):
   ]
   write_report(fn, title=title, keys=sorted_paper_keys, rows=sorted_papers)
 
-def process_paper(fn, addresses, megapixels):
+def process_paper(fn, addresses, megapixels, success):
   res = {
     'paperId': '',
     'key': '',
@@ -177,6 +187,7 @@ def process_paper(fn, addresses, megapixels):
       #   'geocoded': geocoded_institutions,
       # })
       if address:
+        success[citationId] = True
         geocoded_citations.append([
           citation.title,
           institution,
@@ -185,6 +196,7 @@ def process_paper(fn, addresses, megapixels):
           citation.title,
         ] + address)
       else:
+        success[citationId] = False
         unknown_citations.append([
           # citationId,
           citation.title,