Bot conversation designer and tester for in browser conversation flow planning and debugging.
Visit tabahi.github.io/Speako/ to check a very simple bot or design your own.
- There should be at least 1
Bot Start
node in the design. - Each
Bot
andBot Start
should be followed by at least oneHuman
node. - Each
Bot
andBot Start
should be followed by aBot fallback
node for in case no response matches the next availableHuman
nodes. - Adding a star
*
as one of the Text possibilities inHuman
nodes will accept all responses by human, therefore no need for fallback in that case. - The end event nodes
Desired Out
andDrop
can have multiple action commands as Text values. Currently, only the URL command is programmed. - If a
Human
node is followed by multiple bot nodes, then preference goes toDesired Out
, thenBot
and thenDrop
, in that order. If there are multipleBot
nodes available then one will be selected randomly.
- 设计中应该至少有 1 个
Bot Start
节点。 - 每个
Bot
和Bot Start
都应该跟有至少一个Human
节点。 - 每个
Bot
和Bot Start
后面都应该跟一个Bot fallback
节点,以防没有响应匹配下一个可用的Human
节点。 - 添加星号
*
作为Human
节点中的文本可能性之一将接受人类的所有响应,因此在这种情况下不需要回退。 - 结束事件节点
Desired Out
和Drop
可以有多个动作命令作为文本值。 目前,仅对 URL 命令进行编程。 - 如果一个
Human
节点后面跟着多个 bot 节点,则优先级依次为Desired Out
、Bot
和Drop
。 如果有多个“Bot”节点可用,则将随机选择一个。
Designer saves the model in browser's local memory, and then the tester uses that model during the conversation. Closing the browser (usually) won't reset it, however clearing the browser's memory will reset the model.
Currently only Chrome, Edge and Safari has the full support for Speech Recognition. Check MDN docs for the latest news because Web Speech API is currently flagged as experimental technology.
:: Install Webpack
:: cd ./
npm init -y
npm install webpack webpack-cli --save-dev
:: Edit webpack.config.js
npx webpack --config webpack.config.js
:: Build:
npm run build
:: Install and Run Dev server:
npm install webpack-dev-server
npm run dev
Dependency: GoJS 2.1