add dl23 passage/doc and dl22 doc to pyserini (#1804)

This cl integrates dl22 doc and dl23 doc/passages into pyserini TESTED=added load qrel and load query tests
castorini · Mar 5, 2024 · 84c6742 · 84c6742
1 parent 8b21701
commit 84c6742
Show file tree

Hide file tree

Showing 3 changed files with 29 additions and 0 deletions.
diff --git a/pyserini/search/_base.py b/pyserini/search/_base.py
@@ -82,6 +82,7 @@
     'dl22': JTopics.TREC2022_DL,
     'dl22-unicoil': JTopics.TREC2022_DL_UNICOIL,
     'dl22-unicoil-noexp': JTopics.TREC2022_DL_UNICOIL_NOEXP,
+    'dl23': JTopics.TREC2023_DL,
     'msmarco-doc-dev': JTopics.MSMARCO_DOC_DEV,
     'msmarco-doc-dev-unicoil': JTopics.MSMARCO_DOC_DEV_UNICOIL,
     'msmarco-doc-dev-unicoil-noexp': JTopics.MSMARCO_DOC_DEV_UNICOIL_NOEXP,
@@ -407,7 +408,10 @@
     'dl20-passage': JQrels.TREC2020_DL_PASSAGE,
     'dl21-doc': JQrels.TREC2021_DL_DOC,
     'dl21-passage': JQrels.TREC2021_DL_PASSAGE,
+    'dl22-doc': JQrels.TREC2022_DL_DOC,
     'dl22-passage': JQrels.TREC2022_DL_PASSAGE,
+    'dl23-doc': JQrels.TREC2023_DL_DOC,
+    'dl23-passage': JQrels.TREC2023_DL_PASSAGE,
     'msmarco-doc-dev': JQrels.MSMARCO_DOC_DEV,
     'msmarco-passage-dev-subset': JQrels.MSMARCO_PASSAGE_DEV_SUBSET,
     'msmarco-v2-doc-dev': JQrels.MSMARCO_V2_DOC_DEV,

diff --git a/tests/test_load_qrels.py b/tests/test_load_qrels.py
@@ -358,12 +358,31 @@ def test_dl21(self):
         self.assertFalse(isinstance(next(iter(qrels.keys())), str))
 
     def test_dl22(self):
+        qrels = search.get_qrels('dl22-doc')
+        self.assertIsNotNone(qrels)
+        self.assertEqual(len(qrels), 76)
+        self.assertEqual(sum([len(qrels[topic_id]) for topic_id in qrels]), 369638)
+        self.assertFalse(isinstance(next(iter(qrels.keys())), str))
+
         qrels = search.get_qrels('dl22-passage')
         self.assertIsNotNone(qrels)
         self.assertEqual(len(qrels), 76)
         self.assertEqual(sum([len(qrels[topic_id]) for topic_id in qrels]), 386416)
         self.assertFalse(isinstance(next(iter(qrels.keys())), str))
 
+    def test_dl23(self):
+        qrels = search.get_qrels('dl23-doc')
+        self.assertIsNotNone(qrels)
+        self.assertEqual(len(qrels), 82)
+        self.assertEqual(sum([len(qrels[topic_id]) for topic_id in qrels]), 18034)
+        self.assertFalse(isinstance(next(iter(qrels.keys())), str))
+
+        qrels = search.get_qrels('dl23-passage')
+        self.assertIsNotNone(qrels)
+        self.assertEqual(len(qrels), 82)
+        self.assertEqual(sum([len(qrels[topic_id]) for topic_id in qrels]), 22327)
+        self.assertFalse(isinstance(next(iter(qrels.keys())), str))
+
     # Various multi-lingual test collections
     def test_ntcir8_zh(self):
         qrels = search.get_qrels('ntcir8-zh')

diff --git a/tests/test_load_queries.py b/tests/test_load_queries.py
@@ -381,6 +381,12 @@ def test_dl22(self):
         self.assertEqual(len(topics), 500)
         self.assertFalse(isinstance(next(iter(topics.keys())), str))
 
+    def test_dl23(self):
+        topics = search.get_topics('dl23')
+        self.assertIsNotNone(topics)
+        self.assertEqual(len(topics), 700)
+        self.assertFalse(isinstance(next(iter(topics.keys())), str))
+
     # Various multi-lingual test collections
     def test_ntcir8_zh(self):
         topics = search.get_topics('ntcir8-zh')