diff --git a/content/90.back-matter.md b/content/90.back-matter.md
index f660084..a04a5b1 100644
--- a/content/90.back-matter.md
+++ b/content/90.back-matter.md
@@ -1,3 +1,5 @@
+Supplementary information for "The probability of edge existence due to node degree: a baseline for network-based predictions".
+
 ### XSwap parameter settings for network types
 
 | Network type | Degree preserved | Figure | `allow_antiparallel` | `allow_loops` |
@@ -11,7 +13,7 @@ Applications of the modified XSwap algorithm to various network types with appro
 For simple networks, each node's degree is preserved.
 For bipartite networks, each node's number of connections to the other part is preserved, and the partite sets (node class memberships) are preserved.
 For directed networks, each nodes' in- and out-degrees are preserved, though parameter choices depend on the network being permuted.
-Some directed networks can include antiparallel edges or loops while others do not. {#tbl:xswap}
+Some directed networks can include antiparallel edges or loops while others do not. {#tbl:xswap tag="S1"}
 
 ### Performance of the XSwap algorithm
 
@@ -23,7 +25,7 @@ Random graphs generated with a preferential attachment mechanism (via Barabási
 ![
   **Higher density networks have lower asymptotic fractions of edges swapped and take more attempts to reach these values.**
   The Barabási–Albert model produces scale-free random graphs, while Erdős–Rényi generates random graphs where all edges are equally likely.
-](https://github.com/greenelab/xswap-analysis/raw/47f67f85b1a5df2714d564c274515f1fdeb882ba/img/6_xswap_percent_swapped_iterations/lines_continuous.png){#fig:swap-percent width="100%"}
+](https://github.com/greenelab/xswap-analysis/raw/47f67f85b1a5df2714d564c274515f1fdeb882ba/img/6_xswap_percent_swapped_iterations/lines_continuous.png){#fig:swap-percent width="100%" tag="S1"}
 
 <!--
 fig:swap-percent is created by
@@ -278,7 +280,7 @@ All definitions that follow are the score between nodes $u$ and $v$.
 | Random walk with restart score | $c \bigg[ \bigg( \mathbb{I} - (1-c) \mathbf{A}\bigg)^{-1} \mathbf{y}_u \bigg]_v$ | [@doi:10.1145/1014052.1014135;@raw:laplacian] |
 | Inference score | $\frac{|A(u) \cap D(v)|}{|A(u)|} + \frac{|D(u) \cap D(v)|}{|D(u)|}$ | [@doi:10.5821/dissertation-2117-95691] |
 
-Table: Edge prediction features. {#tbl:edge-prediction}
+Table: Edge prediction features. {#tbl:edge-prediction tag="S2"}
 
 ## References {.page_break_before}
 
diff --git a/content/metadata.yaml b/content/metadata.yaml
index ef53e0b..f55c899 100644
--- a/content/metadata.yaml
+++ b/content/metadata.yaml
@@ -1,5 +1,5 @@
 ---
-title: "Supplemental information"
+title: "Supplementary Information"
 keywords:
   - networks
   - heterogeneous