WebSocket API

WebSocket Overview

SymetryML supports the use of WebSocket technology on a subset of its API. This API can be used when lower latency is needed. Each WebSocket call contains a set of 3 pieces of information:

• The header length

• The headers, which are used to authenticate the request.

• The message payload.

The message are string formatted using the following protocol

• <Header-length>,<header><payload> ex:

• 65,{headers=['some date','some md5','some authorization','some customer id'],extraKeys={}}{PAYLOAD}

The following table describes this information.

Header Length	Length of the header section.
Header	A JSON string containing 2 pieces of information - header[4] = an array of 4 string used for authorization. - date - md5 - authorization hash code. - customer id - extraKeys = contains extra information that depends on the WebSocket endpoint invoked. This Header field is mandatory on the first WebSocket message only.
WebSocket Payload	A JSON string representing a dataframe. See the DataFrame for more details

About WebSocket Authorization

With WebSocket, the following string must be used to create the Authorization hash code.

Item	Description
Authorization	Signature
Signature	Base64( HMAC-SHA2(UTF-8-Encoding-Of( StringToSign ) ) );
StringToSign	Content-MD5 + "\n" SymetryML-Secret-Key + "\n" Sym-Date Header + "\n" SymetryML-key-Id + "\n" RequestBody + "\n" ServicePath + "\n" extraKeys
Sym-Date format	See the section Symetry Date Header
ServicePath	Path of the URL to be invoked (for instance /symetry/ws/predict).
extraKeys	Each extra key should be included individually on a separate new line. See example below for details.

Example of WebSocket String to Sign

H47KwjgTU5G7v8BJS+ttgg== + '\n'
<SECRETKEY> + '\n'
2014-10-24 17:24:04 + '\n'
c1 + '\n'
{"attributeNames":["sepal_length","sepal_width","petal_length"],"data":[["4.3","3","1.1"]],"attributeTypes":["C","C","C"]} + '\n'
/symetry/ws/predict + '\n'
irisTest + '\n'
testLDA

Learning Using WebSocket

This API function call allows users to learn a dataframe using WebSocket services.

URL

ws://SERVER_HOST/symetry/ws/learn

Extra Parameter

Index	Short Description	Description
0	SymetryML Project ID	The name of the project to learn the data.

WS Response

HTTP Status Code	HTTP Status Message	Description
200	OK	Success.

Sample WS Request

Request URL:
ws://charm:8080/symetry/ws/learn
Request Message:
{"headers":["2014-10-24 14:28:03","PuAtwbvTrwlX8JQHWvtv7Q\u003d\u003d","Rz/EO7GXX/KE2Jj52+kePmIrudA\u003d","c1"],"extraKeys":["irisTest"]}-289907289{"attributeNames":["sepal_length","sepal_width","petal_length","petal_width","sepal_lengt_b1","sepal_lengt_b2","sepal_width_b1","sepal_width_b2","petal_length_b1","petal_length_b2","petal_width_b1","petal_width_b2","Iris_setosa","Iris_versicolor","Iris_virginica"],"data":[["4.3","3","1.1","0.1","1","0","0","1","1","0","1","0","1","0","0"],["4.8","3","1.4","0.1","1","0","0","1","1","0","1","0","1","0","0"],["4.9","3.1","1.5","0.1","1","0","0","1","1","0","1","0","1","0","0"],["4.9","3.
(...)
5","0","1","0","1","0","1","0","1","0","0","1"],["7.2","3.6","6.1","2.5","0","1","0","1","0","1","0","1","0","0","1"]],"attributeTypes":["C","C","C","C","B","B","B","B","B","B","B","B","B","B","B"]}

Forgetting Using WebSocket

This API function call allows users to forget a DataFrame of data using WebSocket services.

URL

ws://SERVER_HOST/symetry/ws/forget

Extra Parameter

Index	Short Description	Description
0	SymetryML Project ID	The name of the project to learn the data.

WS Response

HTTP Status Code	HTTP Status Message	Description
200	OK	Success.

Predicting Using WebSocket

This API function call allows users to predict using WebSocket services.

URL

ws://SERVER_HOST/symetry/ws/predict

Extra Parameter

Index	Short Description	Description
0	SymetryML Project ID	The name of the project to learn the data.
1	Model Id	The model name to use to make predictions.

WS Response

On success this WebSocket call will return a JSONResponseClient containing a KSVSMap entity. The format of this response is the same as with the SymetryML REST prediction API described in Prediction API.

Sample WS Request

Request URL
ws://charm:8080/symetry/ws/predict

Request Message:
{"headers":["2014-10-24 17:17:58","H47KwjgTU5G7v8BJS+ttgg\u003d\u003d","/Xd/JypB3rDQ+1BVaZRm6n35kt0\u003d","c1"],"extraKeys":["irisTest","testLDA"]}-289907289{"attributeNames":["sepal_length","sepal_width","petal_length"],"data":[["4.3","3","1.1"]],"attributeTypes":["C","C","C"]}

Response:
{"statusCode":"OK","statusString":"OK","values":{"KSVSMap":{"values":[{"res":"1","normZ":"0.2929126883636288","resZ":"-0.4559922893809558","z0":"-3.840891864408894","z1LTz0":"false"}]}}}