diff --git a/Definition-dpp.md b/Definition-dpp.md
new file mode 100644
index 00000000..37ec46e6
--- /dev/null
+++ b/Definition-dpp.md
@@ -0,0 +1,39 @@
+# Definitions
+* A dpp is two hash strings joined by the character ":"
+  We call the first chain chid and the second phid.
+
+* The chid and phid are hash strings of certain values.
+  We call the set of these values Documents.
+  Here we define these values.
+  
+# Chid
+The chid is the part of dpp that defines a device, be it a computer,
+a hard drive, etc. The chid is the most important part of a dpp since
+anyone who comes across a device should be able to play it.
+
+The chid is made up of four values:
+   * type
+   * manufacturer
+   * model
+   * serial_number
+type represents the device type according to the devicehub.
+
+These values are always represented in lowercase.
+These values have to be ordered and concatenated with the character "-"
+So:
+   {type}-{manufacturer}-{model}-{serial_number}
+
+For example:
+```
+HardDrive-Seagate-ST500LT0121DG15-S3P9A81F
+```
+
+In computer types this combination is not perfect and **can lead to collisions**.
+That is why we need a value that is reliable and comes from the manufacturer.
+
+# Phid
+The values of the phid do not have to be reproducible. For this reason, each inventory can establish its own values and its order as a document.
+It is important that each inventory store the document in string so that it can reproduce exactly the document that was hashed. So a document can be verifiable.
+
+In the case of the DeviceHub, we use as the chid document all the values that the Workbench collects that describe the hardware's own data.
+These data change depending on the version of the Workbench used.