impl apidocs descriptions

NOTES.md

@@ -1,11 +1,13 @@
 TODO:
 
-  * push live on https://derwen.ai/docs/txg
-
   * extend GOR to replicate NodePiece/ULTRA ?
 
   * load RDF, to bootstrap/iterate analysis
 
+  * download ZIP: KuzuDB node-link
+  https://neo4j.com/docs/getting-started/data-import/csv-import/
+
+
   * reify GOR, then use FastRP to generate embeddings?
   https://github.com/Knorreman/fastRP
 
@@ -16,9 +18,6 @@ TODO:
   * also eval community detection to condense nodes using k-medoids?
   https://medium.com/neo4j/clustering-graph-data-with-k-medoids-3b6a67ea0873
 
-  * download ZIP: KuzuDB node-link
-  https://neo4j.com/docs/getting-started/data-import/csv-import/
-
   * add conda packaging - but what about OpenNRE ?
   https://conda.github.io/grayskull/
 
@@ -30,5 +29,3 @@ TODO:
   * check out https://github.com/wikipedia2vec/wikipedia2vec
 
   * link `sense2vec` synonyms; make affordances for UI to annotate synonyms
-
-

app.py

@@ -338,8 +338,19 @@
             st.markdown(
                 """
 Using the topological transform given in `lee2023ingram`, construct a
-_graph of relations_ for making enhanced graph inference:
-                """
+_graph of relations_ for enhancing graph inference.
+
+<details>
+  <summary><strong>What does this transform provide?</strong></summary>
+  <p>
+By using a <em>graph of relations</em> dual representation of our graph data, first and foremost we obtain a more compact representation of the relations in the graph, and means of making inferences (e.g., <em>link prediction</em>) where there is substantially more invariance in the training data.
+  </p>
+  <p>
+Also recognize that for a parse graph of a paragraph in the English language, the most interesting nodes will probably be either subjects (<code>nsubj</code>) or direct objects (<code>pobj</code>). Here in the <em>graph of relations</em> we can see illustrated how the important details from <em>entity linking</em> tend to cluster near either <code>nsubj</code> or <code>pobj</code> entities, connected through punctuation. This aspect is not as readily observed in the earlier visualization of the <em>lemma graph</em>.
+  </p>
+</details>
+                """,
+                unsafe_allow_html = True,
             )
 
             start_time = time.time()
@@ -375,7 +386,6 @@
             duration = round(time.time() - start_time, 3)
             st.write(f"transform: {round(duration, 3)} sec, {len(gor.rel_list)} relations")
 
-
             ## download lemma graph
             st.subheader("download the results", divider = "rainbow")
             st.markdown(

docs/ex0_0.md

@@ -538,6 +538,12 @@ pv_graph.show("tmp.fig02.html")
 
 
 
+*What does this transform provide?*
+
+By using a _graph of relations_ dual representation of our graph data, first and foremost we obtain a more compact representation of the relations in the graph, and means of making inferences (e.g., _link prediction_) where there is substantially more invariance in the training data.
+
+Also recognize that for a parse graph of a paragraph in the English language, the most interesting nodes will probably be either subjects (`nsubj`) or direct objects (`pobj`). Here in the _graph of relations_ we see illustrated how the important details from _entity linking_ tend to cluster near either `nsubj` or `pobj` entities, connected through punctuation. This is not as readily observed in the earlier visualization of the _lemma graph_.
+
 ## statistical stack profile instrumentation