VS2N-MongoDB-schemas.md

 # VS2N MongoDB schemas
 
-In VS2N, ...
\ No newline at end of file
+In VS2N, we use MongoDB to store all the collected data from SNN simulations. This makes it possible to deploy VS2N on multiple nodes, for a faster and better user experience especially when working with large networks.
+
+
+Each type of collected data is stored in one collection (the equivalent of a table in a relational database system), and all the collections on one simulation are stored in one MongoDB database.
+
+In VS2N, the types of collected data from each SNN simulation (which are stored as collection each) are:
+
+**Required**
+- General information on the network.
+- Spikes activity in the network.
+- Neurons membrane potential updates in the network.
+- Synapses weight activity in the network.
+
+**Optional**
+- Labels of each input to the network (if possible).
+
+Each collection contains documents (the equivalent of a row in a table of a relational database system), each document has a specific schema that need to be respected by any simulator, in order for VS2N to work properly.
+
+## Document schemas
+A document schema is a JSON object that contains information about the shape, fields and type of data stored in that document.
+
+```JSON
+{
+  "properties" : {
+    "_id" : { "bsonType" : "objectId" },
+    "name" : { "bsonType" : "string" }
+  }
+}
+```
+<center>Example of a document schema</center>
+<br>
+
+## VS2N documents schemas
+## 1- General information schemas:
+In this collection, there are two schemas, the first one contains information about:
+- Used dataset
+- Date of simulation
+- Name of the simulation
+- Number of layers
+- Number of neurons per layer
+
+```JSON
+{ 
+  "properties" : {
+    "_id" : { "bsonType": "objectId" }, // id
+    "n" : { "bsonType": "string" }, // simulation name
+    "L:N" : { "<JSON>" }, // layer name : neurons number
+    "T" : { "bsonType": "string" }, // simulation date
+    "D" : { "bsonType": "string" } // dataset name
+  }
+}
+```
+<center>General information schema - 1</center>
+<br>
+
+```JSON
+{ 
+  "_id" : ObjectId("603834c376e245cd78319032"), 
+  "n" : "My network", 
+  "L:N" : { "Input" : 784, "layer1" : 100 }, 
+  "T" : "26-02-2021-00:37:39", 
+  "D" : "Mnist" 
+}
+```
+<center>Example</center>
+
+The second one contains optional information, which are collected if there was test phase after training:
+- Network accuracy
+- final neuron classes (for non supervised learning)
+
+```JSON
+{ 
+  "properties" : {
+    "_id" : { "bsonType": "objectId" }, // id
+    "MaxS" : { "bsonType": "string" }, // network accuracy (Max Spike)
+    "NLabel" : { "<Array of JSON>" }, // neurons labels
+  }
+}
+```
+<center>General information schema - 2</center>
+<br>
+
+```JSON
+{ 
+  "_id" : ObjectId("60163a04dec986560a8d907a"), 
+  "MaxS" : 83.51,
+  "NLabel" : [ 
+                { "L" : "Layer1", "N" : "13", "Label" : "2" }, 
+                { "L" : "Layer1", "N" : "99", "Label" : "1" },
+                { "L" : "Layer1", "N" : "97", "Label" : "0" },
+                ...
+             ] 
+}
+
+```
+<center>Example</center>
+
+## 2- Spikes activity schemas:
+In this collection, there are two schemas, the first one contains information about:
+- Used dataset
+
+## 3- Neurons membrane potential schemas:
+In this collection, there are two schemas, the first one contains information about:
+- Used dataset
+
+## 4- Synapses weight activity schemas:
+In this collection, there are two schemas, the first one contains information about:
+- Used dataset
\ No newline at end of file