Update blog post ranking for constant tensor tutorial

1 files changed, 60 insertions, 0 deletions

diff --git a/sample-apps/blog-recommendation/src/main/application/searchdefinitions/blog_post.sd b/sample-apps/blog-recommendation/src/main/application/searchdefinitions/blog_post.sd
index 1b92822e425..d7d15045d9c 100644
--- a/sample-apps/blog-recommendation/src/main/application/searchdefinitions/blog_post.sd
+++ b/sample-apps/blog-recommendation/src/main/application/searchdefinitions/blog_post.sd
@@ -63,4 +63,64 @@ search blog_post {
         }
     }
 
+    constant W_hidden {
+        file: constants/W_hidden.json
+        type: tensor(user_item_cf[20],hidden[40]) # user_item_cf
+    }
+
+    constant b_hidden {
+        file: constants/b_hidden.json
+        type: tensor(hidden[40])
+    }
+
+    constant W_final {
+        file: constants/W_final.json
+        type: tensor(hidden[40],prob_success[1])
+    }
+
+    constant b_final {
+        file: constants/b_final.json
+        type: tensor(prob_success[1])
+    }
+
+    # Temporary hack: requires x and y to be 0 padded
+    macro concat(x, y) {
+        expression: x + y
+    }
+
+    # Not valid. Needs to be a predefined function.
+    macro relu(x) {
+        expression: max(0, x) # 0 should be a constant tensor with 0 values
+    }
+
+    # Not valid. Needs to be predefined function.
+    macro sigmoid(x) {
+        expression: 1 / (1 + exp(-x))
+    }
+
+    macro matmul(x, y, dim) {
+        expression: sum(x * y, dim)
+    }
+
+    # The input to the neural network is the concatenation of the query and document vectors
+    macro nn_input() {
+        expression: concat(attribute(user_item_cf), query(user_item_cf))
+    }
+
+    macro hidden_layer() {
+        expression: relu(matmul(nn_input, constant(W_hidden), "user_item_cf") + constant(b_hidden))
+    }
+
+    macro final_layer() {
+        expression: sigmoid(matmul(hidden_layer, constant(W_final), "hidden") + constant(b_final))
+    }
+
+    rank-profile nn_tensor {
+        first-phase {
+            expression {
+                sum(final_layer)
+            }
+        }
+    }
+
 }