test cases and code changes

Author: Gerit Wagner

bib_dedupe/sim.py

@@ -126,7 +126,7 @@ def sim_title(title_1: str, title_2: str, debug: bool = False) -> float:
     t1 = str(title_1)
     t2 = str(title_2)
 
-    if t1 == "" and t2 == "":
+    if t1 in ["", "book review"] or t2 in ["", "book review"]:
         return 0.0
 
     if t1.replace(" ", "") == t2.replace(" ", "") and t1.replace(" ", "") != "":
@@ -140,6 +140,10 @@ def sim_title(title_1: str, title_2: str, debug: bool = False) -> float:
                 "comment",
                 "response",
                 "reply",
+                "update",
+                "forum",
+                "proposed",
+                "talk",
             ]
         ]
     ):

tests/sim_test.py

@@ -149,6 +149,13 @@ def test_sim_container_title(
             "open label, multi center clinical trial eculizumab adult patients atypical hemolytic uremic syndrome",
             0.0,
         ),
+        (
+            "negotiation database schema integration",
+            "",
+            0.0,
+        ),
+        ("proposed study commitment virtual teams", "commitment virtual teams", 0.0),
+        ("task force report", "update of the task force report", 0.0),
         # ("cardiac vascular remodelling effect antihypertensive agents",
         #  "session 2 cardiac vascular remodelling effect antihypertensive agents",
         #  1.0),

tests/test_cases.json

@@ -553,6 +553,73 @@
         "pages": "1--17"
       },
       "expected_duplicate": true
+    },
+    {
+      "id": "ejis_1995_book_review_unknown_vs_specific_book_review",
+      "note": "Book review records: one generic ('Book review' with UNKNOWN author), one specific book review -> should not match without extra book-title logic.",
+      "record_a": {
+        "ENTRYTYPE": "article",
+        "ID": "1",
+        "author": "UNKNOWN",
+        "title": "Book review",
+        "journal": "European Journal of Information Systems",
+        "volume": "4",
+        "number": "2",
+        "year": "1995"
+      },
+      "record_b": {
+        "ENTRYTYPE": "article",
+        "ID": "2",
+        "author": "Hammer, Michael and Champy, James",
+        "title": "Book review: Edited by TONY CORNFORD Reengineering the Corporation: A Manifesto for Business Revolution",
+        "journal": "European Journal of Information Systems",
+        "volume": "4",
+        "number": "2",
+        "year": "1995"
+      },
+      "expected_duplicate": false
+    },
+    {
+      "id": "sekar_2024_literature_review_vs_erf_paper",
+      "note": "LR vs ERF paper; related topic but different titles and authorship -> must NOT match.",
+      "record_a": {
+        "ENTRYTYPE": "inproceedings",
+        "ID": "sekar_tech_noteboom_2024_lit_review",
+        "author": "Sekar, Aravindh and Tech, Deb and Noteboom, Cherie",
+        "year": "2024",
+        "title": "Exploring the Impact of Blockchain Integration on Inventory Accuracy and Supply Chain Efficiency -A Literature Review",
+        "booktitle": "Americas Conference on Information Systems"
+      },
+      "record_b": {
+        "ENTRYTYPE": "inproceedings",
+        "ID": "sekar_tech_2024_erf_paper",
+        "author": "Sekar, Aravindh and Tech, Deb",
+        "year": "2024",
+        "title": "Exploratory Study on the Impact of Blockchain Adoption on Inventory Accuracy and Supply Chain Efficiency Emergent Research Forum (ERF) Paper",
+        "booktitle": "Americas Conference on Information Systems"
+      },
+      "expected_duplicate": false
+    },
+    {
+      "id": "powell_1999_two_distinct_pdfs_similar_authors_year",
+      "note": "Same author/year/venue but different titles -> must NOT match.",
+      "record_a": {
+        "ENTRYTYPE": "inproceedings",
+        "ID": "powell_1999_a_proposed_study",
+        "author": "Powell, Anne L",
+        "year": "1999",
+        "title": "A Proposed Study on Commitment in Virtual Teams",
+        "booktitle": "Americas Conference on Information Systems"
+      },
+      "record_b": {
+        "ENTRYTYPE": "inproceedings",
+        "ID": "powell_1999_commitment_virtual_team",
+        "author": "Powell, Anne L",
+        "year": "1999",
+        "title": "Commitment in a Virtual Team",
+        "booktitle": "Americas Conference on Information Systems"
+      },
+      "expected_duplicate": false
     }
   ]
 }