Distinguish historical primary editions from affiliated workshops

docs/data-sources.md

@@ -178,6 +178,10 @@ paper-atlas collect dblp \
 
 输出按会议/年份/track 分文件，附 manifest、`dblp-normalization.json` 范围审查统计和带哈希的 `dblp-import-queue.json`。不能识别的书目、年份或归属会逐类登记，不默认算作主会。DBLP 不含摘要，因此输出摘要为空、清单为 partial；需要通过有证据的跨来源关联补齐论文实体摘要。书目发表证据与原始审稿决定分别保留，不依据书目缺席推断拒稿。已存在其他出版来源的届次，还须先回填其权威身份并核验镜像记录，不能直接累加投稿数。
 
+早期 AISTATS 1995/1997/1999/2001/2003/2005 和 COLT 1988–1991 的主系列名称本身含 Workshop，不能当作附属 workshop 漏掉。识别要求已核对的完整 proceedings key、会议书目名、原始标题前缀、届数和事件年一致；轨道名为 `main` 表示主系列，保留原始 Workshop 名称及官方目录证据，不改造其历史名称或出版形式。参见 [AISTATS 官方历史目录](https://aistats.org/other.html)、[COLT 协会目录](https://www.learningtheory.org/)。1995 AISTATS 的会议年与1996图书出版年分开；[PMLR R0](https://proceedings.mlr.press/r0/) 是后来重发的 pre-proceedings，不与 DBLP 图书条目盲目逐篇等同。
+
+真正附属的 AISTATS 2010 Active Learning and Experimental Design workshop 仍单列。COLT 2003 与 Kernel Workshop 的 [混合论文集](https://doi.org/10.1007/b12006) 缺少逐篇章节信息，使用 `proceedings_unspecified`，不能整卷改成 main 或附属 workshop。输出保存 `track_scope_note` 与 `track_scope_evidence_url`，导入后仍可追查分类理由；这些出版证据不是真实拒稿或原始决定签名，也不自动修正既有主库记录。
+
 部分早期 SIGIR、CIKM、IJCAI 出版卷使用两位年份键；只有论文集标题明确给出唯一完整年份且与短键相符时才采用，不推测世纪。WWW 特殊兴趣、Alternate track 单列；同卷混合 Applied Data Science、Demo、Industrial 且不能逐篇区分时使用 `proceedings_unspecified`，不把整卷归为某一个具体 track。
 
 书目中的前后页、目录组合和明确会务标题不产生论文录用记录；缺作者的赞助、委员会、活动汇总等标题单独排除。WWW 2019 companion 的匿名合成活动条目归入待核查材料，原始快照仍保留。范围报告分别统计这些原因；单纯缺作者并不排除真实研究论文，也不补造作者。

paper_atlas/collectors/dblp_records.py

@@ -90,6 +90,8 @@ def normalize_row(row,volumes,targets,include_workshops=False,through_year=2026)
               'ambiguous_identifiers':ambiguous,'publication_evidence':'inproceedings bibliography; not an original decision note'}
     if edition.get('track_scope_note'):
         metadata['track_scope_note']=edition['track_scope_note']
+    if edition.get('track_scope_evidence_url'):
+        metadata['track_scope_evidence_url']=edition['track_scope_evidence_url']
     flagged=bool(re.search(r'\b(?:withdrawn|retracted)\b',row.get('publtype') or '',re.I))
     result={'source':'dblp','external_id':row['key'],'url':'https://dblp.org/rec/'+row['key']+'.html',
             'title':title,'abstract':None,'authors':authors,'identifiers':own,

paper_atlas/collectors/dblp_scope.py

@@ -38,6 +38,39 @@ BOOKS = {
 }
 
 
+# These are the historical primary series themselves, not co-located workshops.
+# Exact proceedings keys, original names, ordinals and event years must agree.
+HISTORICAL_WORKSHOP_TITLES = {
+    **{('aistats', year): prefix + ordinal + ' International Workshop on Artificial Intelligence and Statistics'
+       for year, ordinal, prefix in [
+           (1995, 'Fifth', 'Learning from Data - '),
+           (1997, 'Sixth', 'Proceedings of the '),
+           (1999, 'Seventh', 'Proceedings of the '),
+           (2001, 'Eighth', 'Proceedings of the '),
+           (2003, 'Ninth', 'Proceedings of the '),
+           (2005, 'Tenth', 'Proceedings of the ')]},
+    **{('colt', year): 'Proceedings of the ' + ordinal + ' Annual Workshop on Computational Learning Theory'
+       for year, ordinal in [(1988, 'First'), (1989, 'Second'), (1990, 'Third'), (1991, 'Fourth')]},
+}
+
+
+def historical_track(family, base, suffix, volume_key, year, title):
+    if suffix or base.casefold() != family or volume_key != f'conf/{family}/{year}':
+        return None
+    prefix = HISTORICAL_WORKSHOP_TITLES.get((family, year))
+    if (prefix and title.casefold().startswith(prefix.casefold() + ',')
+            and re.search(r'\b' + str(year) + r'\b', title)):
+        url = 'https://aistats.org/other.html' if family == 'aistats' else 'https://www.learningtheory.org/'
+        return ('main', 'Historical primary series named Workshop; exact proceedings key, ordinal, title and event year verified.', url)
+    if (family == 'colt' and year == 2003
+            and title.startswith('Computational Learning Theory and Kernel Machines, 16th Annual Conference on Computational Learning Theory and 7th Kernel Workshop, COLT/Kernel 2003,')
+            and title.endswith('Proceedings')):
+        return ('proceedings_unspecified',
+                'Combined COLT 2003 and Kernel Workshop volume; per-paper section absent from DBLP XML. Do not infer a main or workshop track.',
+                'https://doi.org/10.1007/b12006')
+    return None
+
+
 def scope(row,volume,through_year=2026):
     fields=row['fields'];books=fields.get('booktitle',[])
     if len(books)!=1:return None,'ambiguous_booktitle'
@@ -75,7 +108,9 @@ def scope(row,volume,through_year=2026):
     if len(bibliographic)!=1 or not re.fullmatch(r'\d{4}',bibliographic[0]):return None,'ambiguous_bibliographic_year'
     if int(bibliographic[0])!=year and not re.search(r'\b'+str(year)+r'\b',title):return None,'unverified_year_difference'
     text=(suffix+' '+title).casefold();track='main'
-    if 'findings' in text:track='findings'
+    historical = historical_track(family, base, suffix, volume['key'], year, title)
+    if historical:track=historical[0]
+    elif 'findings' in text:track='findings'
     elif 'shared task' in text:track='shared_task'
     elif 'student research' in text or 'workshop' in text or 'wokshop' in text or base in ('ICCVW','EWSL'):track='workshop'
     elif ('demo track' in text and ('industrial track' in text or 'applied data science track' in text)):track='proceedings_unspecified'
@@ -98,6 +133,9 @@ def scope(row,volume,through_year=2026):
     if venue=='AAMAS' and year<2002:return None,'predecessor_not_aamas'
     result = {'venue':venue,'venue_aliases':aliases,'year':year,'track':track,'bibliographic_year':int(bibliographic[0]),
               'rule':family+':'+base,'proceedings_title':title,'edition_year_basis':year_basis}
+    if historical:
+        result['track_scope_note'] = historical[1]
+        result['track_scope_evidence_url'] = historical[2]
     if venue=='MM' and track=='main':
         # A main MM volume also includes doctoral, tutorials, companion papers
         # and other sections. DBLP XML omits the per-paper TOC section.

tests/test_dblp_records.py

@@ -10,7 +10,7 @@ from unittest.mock import patch
 
 from paper_atlas.collectors.dblp import XML_SHA
 from paper_atlas.collectors.dblp_records import collect,identifiers,normalize_row,material_reason
-from paper_atlas.collectors.dblp_scope import scope
+from paper_atlas.collectors.dblp_scope import scope, HISTORICAL_WORKSHOP_TITLES
 from paper_atlas.db.store import Store
 from paper_atlas.ingestion.records import ingest_file
 
@@ -70,6 +70,69 @@ class DblpRecordTests(unittest.TestCase):
         self.assertEqual(scope(self.row(),self.volume('kdd',2024))[1],'crossref_venue_conflict')
         self.assertEqual(scope(self.row(year=2027),self.volume('aaai',2027))[1],'future_edition')
 
+    def test_historical_primary_series_workshop_names_require_exact_editions(self):
+        self.assertEqual(set(HISTORICAL_WORKSHOP_TITLES),
+                         {('aistats',year) for year in [1995,1997,1999,2001,2003,2005]} |
+                         {('colt',year) for year in [1988,1989,1990,1991]})
+        for (family,year),prefix in HISTORICAL_WORKSHOP_TITLES.items():
+            with self.subTest(family=family,year=year):
+                title=prefix+f', {family.upper()} {year}, Proceedings.'
+                volume=self.volume(family,year,title)
+                row=self.row(family,family.upper(),year)
+                if year==1995:
+                    row['fields']['year']=['1996'];volume['fields']['year']=['1996']
+                edition,reason=scope(row,volume)
+                self.assertIsNone(reason);self.assertEqual(edition['track'],'main')
+                self.assertEqual(edition['year'],year)
+                self.assertIn('Historical primary series',edition['track_scope_note'])
+                record,reason=normalize_row(row,{volume['key']:volume},{family.upper()})
+                self.assertIsNone(reason);self.assertEqual(record['metadata']['track_scope_evidence_url'],edition['track_scope_evidence_url'])
+                self.assertEqual(record['metadata']['proceedings_title'],title)
+                if year==1995:self.assertEqual(record['metadata']['bibliographic_year'],1996)
+
+    def test_historical_exception_does_not_promote_affiliated_or_wrong_scope_workshops(self):
+        title=HISTORICAL_WORKSHOP_TITLES[('aistats',1995)]+', AISTATS 1995, Proceedings.'
+        for book,key,changed in [
+            ('AISTATS','conf/aistats/1995w',title),
+            ('AISTATS (Workshops)','conf/aistats/1995',title),
+            ('AISTATS','conf/aistats/1995',title.replace('Fifth','Sixth')),
+            ('AISTATS','conf/aistats/1995',title.replace('1995','1997'))]:
+            volume=self.volume('aistats',1995,changed);volume['key']=key
+            row=self.row('aistats',book,1995);row['fields']['crossref']=[key]
+            self.assertEqual(scope(row,volume)[0]['track'],'workshop')
+        volume=self.volume('aistats',2010,'Active Learning and Experimental Design workshop, In conjunction with AISTATS 2010')
+        volume['key']='conf/aistats/2010w'
+        row=self.row('aistats','AISTATS',2011);row['fields']['crossref']=[volume['key']]
+        self.assertEqual(scope(row,volume)[0]['track'],'workshop')
+
+    def test_combined_colt_kernel_volume_keeps_unknown_per_paper_track(self):
+        title='Computational Learning Theory and Kernel Machines, 16th Annual Conference on Computational Learning Theory and 7th Kernel Workshop, COLT/Kernel 2003, Washington, DC, USA, August 24-27, 2003, Proceedings'
+        volume=self.volume('colt',2003,title);row=self.row('colt','COLT',2003)
+        record,reason=normalize_row(row,{volume['key']:volume},{'COLT'})
+        self.assertIsNone(reason);self.assertEqual(record['track'],'proceedings_unspecified')
+        self.assertEqual(record['status'],'accepted')
+        self.assertIn('per-paper section absent',record['metadata']['track_scope_note'])
+        self.assertEqual(record['metadata']['track_scope_evidence_url'],'https://doi.org/10.1007/b12006')
+        volume['key']='conf/colt/2003w';row['fields']['crossref']=[volume['key']]
+        self.assertEqual(scope(row,volume)[0]['track'],'workshop')
+
+    def test_historical_scope_provenance_survives_file_ingestion(self):
+        title=HISTORICAL_WORKSHOP_TITLES[('colt',1988)]+', COLT 1988, Proceedings.'
+        row=self.row('colt','COLT',1988);volume=self.volume('colt',1988,title)
+        record,reason=normalize_row(row,{volume['key']:volume},{'COLT'})
+        self.assertIsNone(reason)
+        with tempfile.TemporaryDirectory() as tmp:
+            root=Path(tmp);path=root/'history.jsonl'
+            path.write_text(json.dumps(record)+'\n')
+            path.with_suffix('.manifest.json').write_text(json.dumps({'source':'dblp','count':1,'list_url':'https://doi.org/10.4230/dblp.xml.2026-11-01','file_sha256':hashlib.sha256(path.read_bytes()).hexdigest()}))
+            ingest_file(root/'sample.sqlite',path)
+            with closing(sqlite3.connect(root/'sample.sqlite')) as connection:
+                track=connection.execute('SELECT track FROM editions').fetchone()[0]
+                metadata=json.loads(connection.execute('SELECT metadata_json FROM source_records').fetchone()[0])
+            self.assertEqual(track,'main')
+            self.assertEqual(metadata['track_scope_evidence_url'],'https://www.learningtheory.org/')
+            self.assertEqual(metadata['proceedings_title'],title)
+
     def test_only_own_identifier_links_are_used_and_ambiguity_is_preserved(self):
         own,ambiguous=identifiers({'ee':['https://doi.org/10.1234/published','https://doi.org/10.48550/arXiv.2401.00001','https://openreview.net/forum?id=realForum'],
                                     'cite':['https://doi.org/10.9999/reference']})