The Inbox module allows you (or your bot trainers) to validate the work of the bot, to teach him new things to make him smarter. In the AI jargon, this task is called "Supervised Learning"
The inbox is basically a list of all the incoming messages users have sent to your bot.
Then you have FILTERS and ACTIONS to find and make use of the messages to improve your bot.
Each message comes with a set of fields:
Occurences: The number of times this message was sent to the bot. May be a good indication of the importance of taking action on the said message.
Message: The user message
Intent: The intent found for the message. May be empty if the message was not understood. May also be wrong if the bot has been confused with another similar intent.
Confidence: The score attached to the intent detection of the bot, of course, the higher is the best.
Time: The date and time where the message was received. Time can be seen on mouse on hover of the date.
Each message also comes with a green / orange / green light as shown below
-Green: the message is on the training dataset, and was understood
All good, nothing to do there.
-Orange: the message was understood but the message is not in the training dataset.
If the bot was right, you can approve the interpretation , the new message will be added to your dataset. If the bot was wrong, you can fix this by attaching this message to another intent.
-Red: the message was not understood
You have two options there:
- Attach the message to the appropriate intent, either an existing or a new one
- Archive the message in a waiting box so you can manage it later
We hope that, at one point, your bot will have much much success, then you may use our filters to target some specific messages.
Here is the current available filters.
1/ Search bar: the search box let you search for specific words in user messages
2/ Language: the language version of your bot ( can be different of the true language used by the user)
3/ Channel: the channel used by the user to send his messages
4/Status: let you focus on the status of the message in the platform.
Here is the current available options in the status filter:
All status: will give you all the messages
Understood: will give access to all the understood messages
Not understood: will give you access to all the messages not understood by the bot
Learned: will give you access to all the messages you have approved or allocated to an intent
Here is for example the list of the misunderstood messages for the last 7 days
5/Archived: let you see the messages you have archived for future analysis
6/Date: let you define the specific period of time of the search.
By default, this filter is set with the last 7 days.
Once you have narrowed your search with the filters, you can select the messages you want to address by clicking on them. The selected messages will have have a specific color.
Once you have selected the messages you want to act on, here is your options:
only add clean data to your model
Although it may be tempting to add as much as possible to your data it can have a bad side effect because you may add noise or unclean data to your model.
👉 Prefer short sentences than long ones with many intents on a single phrase
👉 Fix typos before adding a sentence to your model, it will help the typo correction module stay efficient
Approving a prediction is a way to confirm to the bot that he has done a good job on the selected messages and he can add them to their respective intents if they aren't already present.
Also, approving a message will set his confidence score to 100%.
If some messages are out of scope or you want to address them later, you can archive them.
When archived, messages wont appear in your search results unless you select the
Archived value in the Archive filter
If the bot has mistakenly associated a message with an intent, you may want to fix that with the
Add to an existing intent or
Add to a new existing intent.
In both cases the sentence will be removed from the source intent to be moved to the target intent.
Don't forget to train your bot 😉
Once the supervision is done, you may ask the admin or a developer to train the bot so the changes you made become effective in your bot
Updated 4 months ago