Semantic Summary

Voice interaction flow chart

Semantic - Intelligent semantic open system

Semantic, it is the intelligent semantic open system was developed by Sanbot Cloud, combined with the natural language processing, information retrieval, deep learning and other leading technology, provide a set of semantic parsing, keyword extraction, intelligent questions and answers in the integration of technology to access system, the system provides for the developers to provide complete hardware and software solutions, provide resources for mass general q&a, custom through industry, covering more than vertical semantic general scenario, meet the demand of user customization.

Six major advantages

1. Depth customization

A public semantic library is provided to customize the industry specific voice package for a particular industry.

2. Private cloud deployment

Special industries can deploy their own private cloud services, exclusive of all voice data, can be used in pure LAN.

3. Noise processing module

The hard noise such as echo, current and sound is processed again, and the interference noise is eliminated. The unique soft noise processing module is very perfect for word recognition and word reduction to restore the true semantics.

4. Synonym processing

Large public synonyms, no worries, and support for the private establishment of synonyms.

5. Software + hardware, overall solution output

The open source semantic API + the Sanbot SDK, the user development threshold is low, eliminates development costs from scratch, and provides complete developer technical support.

6. Compatible with multiple semantic schemes

Parallel access Xunfei, Baidu, IBM Watson, Nuance and many other semantic scheme, users according to their own preferences and semantic results.

Companion semantics API

1. Entry semantic in the web page


Enter and display semantic question and answer messages.

WEB entry tool URL:

Menu: Semantic management -> Choose Categories(Creating Categories) -> Batch Import/Single Import (Need to login)

Input method:

single question and answer message entry or CSV file batch entry
(1). A single question and answer message entry
Click on the single question and answer message button, fill in the questions and answers, and then click the Add button;
(2). CSV file batch entry
Click the CSV file batch button, as required to upload your CSV file, and then click the Add button ;


(1). A single question and answer message entry
If the entry is successful, the page will add a successful message, if the entry fails, the page will prompt the corresponding error message;
(2). CSV file batch entry
If all the entries will be successful , the page will increase all successful messages, if some of the entry fails, it will display a list of error messages page, the user can view the reasons for failure,when you view finished ,click on the left navigation bar "Add semantics" to back to the semantics page.


Entry semantic in the web page example.csv Click to download (Need to login)

2. Semantic Recognition API


Fetching resultof semantic recognition

HTTP request method: GET


Request URL:

Request parameters has two options: MUST and OPTIONAL;

Request parameters:
Name Type Option Comment
semantics string Must Content for semantic recognition
token string Must token
Name Type Comment
rc int Error code
text string Source for semantic recognition
precision double Precision of recognition
ans string Result of semantic recognition
classification int Type of returned result:
=0 refers to standard text content;
=1 refersto robot control command;
reason string Error Description:
When rc is not equal to 0, it is an error;
When rc is equal to 0, this field is not returned;
Error Code Description (rc code):
rc Description
0 Success
-1 Internal server error
-2 Sever doesn’t understand or cannot handle passed content
-3 Parameter error (Invalid semantics or token)
-4 Length of semantics or token exceeds limitation;
Maximum size for semantics is 2048 bytes;
Maximum size for token is 40 bytes;
-5 Invalid token
Request Example: province is Shenzhen located in&token=1111111111111111111
"rc": 0,
"text": "What province is Shenzhen located in",
"precision": 0.8887,
"classification": 0,
"ans": "Guangdong"