update metadata compare

src/analyzers/metadata_comparator.py

@@ -74,7 +74,7 @@ class MetadataComparator:
         title_match = title_similarity >= self.TITLE_THRESHOLD
         
         if not title_match:
-            issues.append(f"Title mismatch (similarity: {title_similarity:.2%})")
+            issues.append(f"Title mismatch. Bib: '{bib_entry.title}', Retrieved: '{arxiv_meta.title}'")
         
         # Compare authors
         bib_authors = self.normalizer.normalize_author_list(bib_entry.author)
@@ -84,7 +84,7 @@ class MetadataComparator:
         author_match = author_similarity >= self.AUTHOR_THRESHOLD
         
         if not author_match:
-            issues.append(f"Author mismatch (similarity: {author_similarity:.2%})")
+            issues.append(f"Author mismatch. Bib: {', '.join(bib_authors)}, Retrieved: {', '.join(arxiv_authors)}")
         
         # Compare years
         bib_year = bib_entry.year.strip()
@@ -92,7 +92,7 @@ class MetadataComparator:
         year_match = bib_year == arxiv_year
         
         if not year_match and bib_year and arxiv_year:
-            issues.append(f"Year mismatch: bib={bib_year}, arxiv={arxiv_year}")
+            issues.append(f"Year mismatch. Bib: {bib_year}, Retrieved: {arxiv_year}")
         
         # Overall assessment
         is_match = title_match and author_match
@@ -133,7 +133,7 @@ class MetadataComparator:
         title_match = title_similarity >= self.TITLE_THRESHOLD
         
         if not title_match:
-            issues.append(f"Title mismatch (similarity: {title_similarity:.2%})")
+            issues.append(f"Title mismatch. Bib: '{bib_entry.title}', Retrieved: '{scholar_result.title}'")
         
         # Compare authors (Scholar format is less structured)
         bib_authors = self.normalizer.normalize_author_list(bib_entry.author)
@@ -145,7 +145,7 @@ class MetadataComparator:
         author_match = author_similarity >= self.AUTHOR_THRESHOLD
         
         if not author_match:
-            issues.append(f"Author mismatch (similarity: {author_similarity:.2%})")
+            issues.append(f"Author mismatch. Bib: {', '.join(bib_authors)}, Retrieved: {', '.join(scholar_authors)}")
         
         # Compare years
         bib_year = bib_entry.year.strip()
@@ -153,7 +153,7 @@ class MetadataComparator:
         year_match = bib_year == scholar_year
         
         if not year_match and bib_year and scholar_year:
-            issues.append(f"Year mismatch: bib={bib_year}, scholar={scholar_year}")
+            issues.append(f"Year mismatch. Bib: {bib_year}, Retrieved: {scholar_year}")
         
         # Overall assessment
         is_match = title_match and author_match
@@ -194,7 +194,7 @@ class MetadataComparator:
         title_match = title_similarity >= self.TITLE_THRESHOLD
         
         if not title_match:
-            issues.append(f"Title mismatch (similarity: {title_similarity:.2%})")
+            issues.append(f"Title mismatch. Bib: '{bib_entry.title}', Retrieved: '{crossref_result.title}'")
         
         # Compare authors
         bib_authors = self.normalizer.normalize_author_list(bib_entry.author)
@@ -204,7 +204,7 @@ class MetadataComparator:
         author_match = author_similarity >= self.AUTHOR_THRESHOLD
         
         if not author_match:
-            issues.append(f"Author mismatch (similarity: {author_similarity:.2%})")
+            issues.append(f"Author mismatch. Bib: {', '.join(bib_authors)}, Retrieved: {', '.join(crossref_authors)}")
         
         # Compare years
         bib_year = bib_entry.year.strip()
@@ -212,7 +212,7 @@ class MetadataComparator:
         year_match = bib_year == crossref_year
         
         if not year_match and bib_year and crossref_year:
-            issues.append(f"Year mismatch: bib={bib_year}, crossref={crossref_year}")
+            issues.append(f"Year mismatch. Bib: {bib_year}, Retrieved: {crossref_year}")
         
         # Overall assessment
         is_match = title_match and author_match
@@ -312,7 +312,7 @@ class MetadataComparator:
         title_match = title_similarity >= self.TITLE_THRESHOLD
         
         if not title_match:
-            issues.append(f"Title mismatch (similarity: {title_similarity:.2%})")
+            issues.append(f"Title mismatch. Bib: '{bib_entry.title}', Retrieved: '{ss_result.title}'")
         
         # Compare authors
         bib_authors = self.normalizer.normalize_author_list(bib_entry.author)
@@ -322,7 +322,7 @@ class MetadataComparator:
         author_match = author_similarity >= self.AUTHOR_THRESHOLD
         
         if not author_match:
-            issues.append(f"Author mismatch (similarity: {author_similarity:.2%})")
+            issues.append(f"Author mismatch. Bib: {', '.join(bib_authors)}, Retrieved: {', '.join(ss_authors)}")
         
         # Compare years
         bib_year = bib_entry.year.strip()
@@ -330,7 +330,7 @@ class MetadataComparator:
         year_match = bib_year == ss_year
         
         if not year_match and bib_year and ss_year:
-            issues.append(f"Year mismatch: bib={bib_year}, semantic_scholar={ss_year}")
+            issues.append(f"Year mismatch. Bib: {bib_year}, Retrieved: {ss_year}")
         
         # Overall assessment
         is_match = title_match and author_match
@@ -371,7 +371,7 @@ class MetadataComparator:
         title_match = title_similarity >= self.TITLE_THRESHOLD
         
         if not title_match:
-            issues.append(f"Title mismatch (similarity: {title_similarity:.2%})")
+            issues.append(f"Title mismatch. Bib: '{bib_entry.title}', Retrieved: '{oa_result.title}'")
         
         # Compare authors
         bib_authors = self.normalizer.normalize_author_list(bib_entry.author)
@@ -381,7 +381,7 @@ class MetadataComparator:
         author_match = author_similarity >= self.AUTHOR_THRESHOLD
         
         if not author_match:
-            issues.append(f"Author mismatch (similarity: {author_similarity:.2%})")
+            issues.append(f"Author mismatch. Bib: {', '.join(bib_authors)}, Retrieved: {', '.join(oa_authors)}")
         
         # Compare years
         bib_year = bib_entry.year.strip()
@@ -389,7 +389,7 @@ class MetadataComparator:
         year_match = bib_year == oa_year
         
         if not year_match and bib_year and oa_year:
-            issues.append(f"Year mismatch: bib={bib_year}, openalex={oa_year}")
+            issues.append(f"Year mismatch. Bib: {bib_year}, Retrieved: {oa_year}")
         
         # Overall assessment
         is_match = title_match and author_match
@@ -430,7 +430,7 @@ class MetadataComparator:
         title_match = title_similarity >= self.TITLE_THRESHOLD
         
         if not title_match:
-            issues.append(f"Title mismatch (similarity: {title_similarity:.2%})")
+            issues.append(f"Title mismatch. Bib: '{bib_entry.title}', Retrieved: '{dblp_result.title}'")
         
         # Compare authors
         bib_authors = self.normalizer.normalize_author_list(bib_entry.author)
@@ -440,7 +440,7 @@ class MetadataComparator:
         author_match = author_similarity >= self.AUTHOR_THRESHOLD
         
         if not author_match:
-            issues.append(f"Author mismatch (similarity: {author_similarity:.2%})")
+            issues.append(f"Author mismatch. Bib: {', '.join(bib_authors)}, Retrieved: {', '.join(dblp_authors)}")
         
         # Compare years
         bib_year = bib_entry.year.strip()
@@ -448,7 +448,7 @@ class MetadataComparator:
         year_match = bib_year == dblp_year
         
         if not year_match and bib_year and dblp_year:
-            issues.append(f"Year mismatch: bib={bib_year}, dblp={dblp_year}")
+            issues.append(f"Year mismatch. Bib: {bib_year}, Retrieved: {dblp_year}")
         
         # Overall assessment
         is_match = title_match and author_match