Creating Your Bot
- Owner - The user who created a bot
- Query - A statement issued by either the user or the bot
- Corpus - a bot's body of knowledge, used for both interpreting an incoming query and as a source for responding with an answer
- Bot Name - Give your bot a name, e.g. My Cool Chatbot
- Description - Describe your bot, e.g. I like to talk about unicorns and ponies
- Path - This will get filled in for you by default. You can change it if you prefer, but if in doubt leave it as is.
- Theme - Choose a theme for how your bot will look
- Status - Whether your bot is currently on or off. Turn it off to prevent people from chatting with it. When your bot is off, only you (the owner) can chat with it.
- Initial Dialog - Whether the first statement in a chat session should come from the bot or from the user interacting with it.
- Retain Conversations From - Who your bot should retain conversations from: All (everybody), Owner (just you), Public (everybody but you), or Nobody. This setting affects the Corpus Sources setting (below).
- Corpus Sources - Where does your bot's body of knowledge (corpus) come from: All (both training and conversations), Training (only from training provided by its owner), or Conversations (only from conversations)
- Conversation Corpus Sources - If Corpus Sources is set to either All or Conversations, this setting determines which conversations contribute to your bot's corpus: All (conversations from both the bot's owner and the public), Owner (only conversations with the bot's owner), or Public (only conversations with the public).
- Clone Dialog From - Which bot to seed your bot's corpus with initially. (Currently there are no bots configured to be corpus seeds.)
- API Status - Whether your bot will respond to API requests.
- Minimum Score Threshold - The minimum score needed for a query entered by the user to match the bot's corpus. 0 = no minimum threshold (a user query will match something in the corpus, even if it's not a close match) 1 = high threshold (a user query will only match something in the corpus if it's close to an exact match).
For example, if the bot's corpus contains the phrase "How are you?" and the user's query is "How r u?", a minimum score threshold of 0 is likely to cause that to match (depending on the bot's algorithm settings; see below) but if the minimum score threshold is 1, "How r u?" will likely not match.
- Banned Phrases - a list of phrases which your bot will not respond to. Phrases are case-insensitive, and should be entered one per line. For example:
poopPhrases can consist of single words or multiple words. User queries containing any of these phrases (case-insensitive) will be ignored by your bot. The user will receive a message ("Banned phrase") indicating that they entered a banned phrase. Even if the bot is configured to retain conversations, the user's response will be ignored by the bot and will not be recorded.
foolish human being
If a phrase is banned after it has entered the bot's corpus, it is still considered part of the corpus and may be used to generate a response to a non-banned query. If you want to remove a phrase from the bot's corpus so it is never used as a response, you must edit the corpus directly.
- Banned Phrase Error - a custom message to display to users when they enter a banned phrase. If left blank, the user will see the default message ("Banned phrase").
- Algorithm Weighting - How much weight to give to each algorithm:
0 = no weight (algorithm is ignored)
1 = full weight
For more information see the Algorithms page.