diff --git a/entries/ogbl_wikikg2/__init__.py b/entries/ogbl_wikikg2/__init__.py
new file mode 100644
index 0000000..0e67c60
--- /dev/null
+++ b/entries/ogbl_wikikg2/__init__.py
@@ -0,0 +1,102 @@
+#! /usr/bin/env python
+# -*- coding: utf-8 -*-
+#
+# Copyright (C) 2026 Tiago de Paula Peixoto <tiago@skewed.de>
+#
+# This program is free software: you can redistribute it and/or modify it under
+# the terms of the GNU Affero General Public License as published by the Free
+# Software Foundation, either version 3 of the License, or (at your option) any
+# later version.
+#
+# This program is distributed in the hope that it will be useful, but WITHOUT
+# ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
+# FOR A PARTICULAR PURPOSE.  See the GNU Affero General Public License for more
+# details.
+#
+# You should have received a copy of the GNU Affero General Public License along
+# with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+
+
+import pandas
+
+from .. import *
+
+
+title = "Wikidata knowledge graph (ogbl-wikikg2, 2015)"
+description = """A knowledge graph extracted from Wikidata, from the `ogbl-wikikg2` dataset of the Open Graph Benchmark (OGB), a benchmark for link prediction (version 1, of December 2020). Each node is an entity of Wikidata (e.g. a person, a place, a work or a concept), and a directed edge from entity A to entity B is a statement (A, relation, B) of Wikidata with one of 535 relations (Wikidata properties, e.g. `P31`, "instance of", or `P27`, "country of citizenship"). Pairs of entities may be connected by several edges with different relations. Rare entities were filtered out by the OGB authors.
+
+The triples come from three snapshots of Wikidata of 2015: those of May 2015 form the training set of the benchmark, and those added in the snapshots of August and November 2015 form the validation and test sets (429,456 and 598,543 triples); all of them are included, and the set is given as an edge property. The negative samples of the benchmark are left out. The data do not include the labels of the entities and relations, which can be looked up in Wikidata (some entities and relations may have been removed from Wikidata since 2015).
+
+The network has the following properties:
+
+- `wikidata_id`: the Wikidata identifier of the entity, e.g. `Q92894`.
+- Edge `relation`: the Wikidata property of the relation, e.g. `P31`.
+- Edge `split`: the set of the benchmark the triple belongs to: `train` (May 2015), `valid` (August 2015) or `test` (November 2015)."""
+tags = ['Informational', 'Relatedness', 'Unweighted', 'Multigraph', 'Metadata']
+url = ['https://ogb.stanford.edu/docs/linkprop/#ogbl-wikikg2',
+       'https://snap.stanford.edu/ogb/data/']
+citation = [('W. Hu, M. Fey, M. Zitnik, Y. Dong, H. Ren, B. Liu, M. Catasta, J. Leskovec, "Open Graph Benchmark: Datasets for Machine Learning on Graphs." Advances in Neural Information Processing Systems 33, 22118-22133 (2020)', 'https://doi.org/10.48550/arXiv.2005.00687'),
+            ('D. Vrandečić, M. Krötzsch, "Wikidata: a free collaborative knowledgebase." Communications of the ACM 57(10), 78-85 (2014)', 'https://doi.org/10.1145/2629489')]
+icon_hash = None
+upstream_license = ('CC0 1.0', 'https://creativecommons.org/publicdomain/zero/1.0/')
+upstream_prefix = 'https://snap.stanford.edu/ogb/data'
+z = 'linkproppred/wikikg-v2.zip:wikikg-v2'
+splits = ['train', 'valid', 'test']
+files = [((f'{z}/mapping/nodeidx2entityid.csv.gz',) +
+          tuple(f'{z}/original/{s}_2015.csv.gz' for s in splits), None, None)]
+
+def fetch_upstream(force=False):
+    return fetch_upstream_files(__name__.split(".")[-1], upstream_prefix, files,
+                                force)
+
+def read_csv_gz(name, fname, **kwargs):
+    with open_upstream_file(name, fname, "rb") as f:
+        return pandas.read_csv(f, compression="gzip", **kwargs)
+
+@cache_network()
+@coerce_props()
+@annotate()
+def parse(alts=None):
+    global files
+    name = __name__.split(".")[-1]
+    for fnames, alt, fmt in files:
+        if alts is not None and alt not in alts:
+            continue
+        # "node idx,entity id"
+        nodes = read_csv_gz(name, fnames[0])
+        assert (nodes["node idx"].values == numpy.arange(len(nodes))).all()
+        entities = pandas.Index(nodes["entity id"].values)
+        del nodes
+        assert entities.is_unique
+
+        g = Graph(directed=True)
+        g.add_vertex(len(entities))
+        rels = []
+        esplits = []
+        extra = []
+        for fname, split in zip(fnames[1:], splits):
+            # "<head>,<relation>,<tail>"
+            t = read_csv_gz(name, fname, header=None, dtype="category")
+            ends = []
+            for c in [0, 2]:
+                idx = entities.get_indexer(t[c].cat.categories)
+                missing = idx < 0
+                if missing.any():
+                    # entities absent from the mapping (none are expected)
+                    new = t[c].cat.categories[missing]
+                    extra.extend(new)
+                    entities = entities.append(pandas.Index(new))
+                    idx = entities.get_indexer(t[c].cat.categories)
+                ends.append(idx[t[c].cat.codes.values])
+            g.add_edge_list(numpy.column_stack(ends))
+            cats = list(t[1].cat.categories)
+            rels.extend(cats[i] for i in t[1].cat.codes.values)
+            esplits.extend([split] * len(t))
+            del t, ends
+        assert len(extra) == 0, extra[:10]
+        g.vp.wikidata_id = g.new_vp("string", vals=entities)
+        g.ep.relation = g.new_ep("string", vals=rels)
+        del rels
+        g.ep.split = g.new_ep("string", vals=esplits)
+        yield alt, g