[Example] Fix ACM data JSON content inconsistency (#1106)

*Issue #, if available:* *Description of changes:* This PR fixes the inconsistency of ACM data's config.json for using a dummy categorical edge feature on the paper-citing-paper edge type. And update the corresponding tutorial contents. Preview page of the updated tutorial: https://james4graphstorm.readthedocs.io/en/james_acm_fix_json/tutorials/own-data.html By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice. Co-authored-by: Ubuntu <[email protected]>
awslabs · Dec 4, 2024 · 933f6e4 · 933f6e4
1 parent 618c3fc
commit 933f6e4
Show file tree

Hide file tree

Showing 3 changed files with 14 additions and 1 deletion.
diff --git a/docs/source/tutorials/own-data.rst b/docs/source/tutorials/own-data.rst
@@ -126,6 +126,15 @@ The above command automatically creates the examplary ACM `config.json` file, so
                 ],
                 "source_id_col": "source_id",
                 "dest_id_col": "dest_id",
+                "features": [
+                    {
+                        "feature_col": "cate_feat",
+                        "feature_name": "cate_feat",
+                        "transform": {
+                            "name": "to_categorical"
+                        }
+                    }
+                ],
                 "labels": [
                     {
                         "task_type": "link_prediction",

diff --git a/examples/acm_data.py b/examples/acm_data.py
@@ -257,10 +257,14 @@ def create_acm_raw_data(graph,
                                                                 # The split pct values are just for
                                                                 # demonstration purpose.
                 labels_list.append(label_dict)
+            elif col.startswith('cate_'):                       # Dummy categorical features that ask
+                feat_dict['feature_col'] = col                  # for a "to_categorical" tranformation
+                feat_dict['feature_name'] = col                 # operation
+                feat_dict['transform'] = {"name": "to_categorical"}
+                feats_list.append(feat_dict)
             else:
                 feat_dict['feature_col'] = col
                 feat_dict['feature_name'] = col
-                # for this example, we do not have transform for features
                 feats_list.append(feat_dict)
         # set up the rest fileds of this node type
         if feats_list:

diff --git a/tutorial/ACM_raw_parquet.png b/tutorial/ACM_raw_parquet.png