Early vs full article (#50)

* cases and match-condition (#49)

* case and match condition

Author: geritwagner

bib_dedupe/match_conditions.py

@@ -92,6 +92,13 @@ def both_entrytypes(entrytype: str) -> str:
     f"(({match(DOI)} & ~(doi_1 == '' | doi_2 == '')) & ({TITLE} > 0.95) & ({AUTHOR} > 0.9) & ({YEAR} > 0.9)) & {non_contradicting(CONTAINER_TITLE)} ",
     # no TITLE
     f"({au10_tiXX_ct10} & {match(VOLUME, NUMBER, PAGES, YEAR)} & {non_contradicting(DOI)} & ({ABSTRACT} > 0.95 | {non_contradicting(ABSTRACT)}))",  # typically for number-mismatches in title
+    # early_view_vs_final
+    f"({au095_ti09_ct075}"
+    f" & {non_contradicting(DOI)}"
+    f" & ((volume_1 != '' & volume_2 == '') | (volume_2 != '' & volume_1 == ''))"
+    f" & ((number_1 != '' & number_2 == '') | (number_2 != '' & number_1 == '') | {non_contradicting(NUMBER)})"
+    f" & (pages_1.str.match('^1[-–]') | pages_2.str.match('^1[-–]'))"
+    f")",
 ]
 
 non_duplicate_conditions = [

tests/test_cases.json

@@ -527,6 +527,32 @@
         "booktitle": "Americas Conference on Information Systems"
       },
       "expected_duplicate": false
+    },
+    {
+      "id": "pare_wagner_prester_jds_2024_vs_2023_online_first",
+      "note": "Likely same article; year/pages differ (online-first vs issue); record_a includes DOI.",
+      "record_a": {
+        "ENTRYTYPE": "article",
+        "ID": "1",
+        "author": "Paré, Guy and Wagner, Gerit and Prester, Julian",
+        "title": "How to develop and frame impactful review articles: key recommendations",
+        "journal": "Journal of Decision Systems",
+        "year": "2024",
+        "volume": "33",
+        "number": "4",
+        "pages": "566--582",
+        "doi": "10.1080/12460125.2023.2197701"
+      },
+      "record_b": {
+        "ENTRYTYPE": "article",
+        "ID": "2",
+        "author": "Paré, Guy and Wagner, Gerit and Prester, Julian",
+        "title": "How to develop and frame impactful review articles: key recommendations",
+        "journal": "Journal of Decision Systems",
+        "year": "2023",
+        "pages": "1--17"
+      },
+      "expected_duplicate": true
     }
   ]
 }