As with any regression testing, changes to code and functionality can affect previous results. That is the entire point of regression testing. If changes never affected existing results, then there would be no point to regression testing.

Thanks for your answer.

Bot Libre's NLP does not use deep learning, but a similar machine learning algorithm that allows the bot to analyse the words in a user's input and match it to one of its trained questions and response.

I understand that Bot works based on script and some additional intelligence applied to find related answer for the user question. Can I know which specific aspect of BotLibre is based on deep learning? 

Even without keywords, adding new response will affect what response a bot chooses, if the new response better matches a question by the user.

A response will always be used for the question it was add for, but for similar questions the bot will always try to find the best matching response. To have a response used for many variations of a question, define an intent label for the response and add it as the response for the valid variations.

Hi,

Thanks for quick reply.

I see 2 changes in your suggestion.  1) Additional question with "policy" word added and 2) set "life" as keyword (which should be in both question I think).

With these changes, now it will work fine. However, my point is how do I avoid such regression?

Earlier I had just first 2 questions in script file. Without having question  "what is insurance policy?" in script file, below were answers

Note that, "what is life insurance policy" equally matches "what is life insurance" and "what is insurance policy", so even without the keyword, the bot may use either response.

I would recommend you also add "life" as a keyword to the first response.

Also, if you want "what is life insurance policy" to use the first response then add it as a valid question for this label as well.

Though mentioning keywords is very useful I find very dangerous too.  e.g. I have 2 questions  

what is life insurance?
this is answer for life insurance

what is standard cover in policy?
answer for standard cover
keywords: standard

This works fine and give answer if someone ask "what is life insurance policy". (Assume I didn't consider adding policy as word in question).

Which SDK are you using, the JavaScript SDK?

Please include your example code.

I choose a 3d avatar from my bot and get the id then include it on my sdk and build. Same process with my first avatar as stated on tutorial.

Are you using the avatar in your bot?

or are you using the avatar directly from JavaScript?

For example, i search Julie as my avatar, and get the id and embed it in sdk, how am i going to used Julie as my avatar?

To have your avatar display emotions it requires media (video, image) tagged with the emotion, action, or pose. Make sure to only tag a media with one, as if you tag a media with "happy" and "smile" it will only use that media when it is happy AND smiles.

Not all the avatars in our avatar library support emotions, query the "emotions" tag for avatars that support emotions.

I did this steps but unfortunately  it did not work, no facial  expressions  of emotions  has shown on avatar, it only shows the action like happy, sad, etc on the screen. What i want is gestures and emotions . 

Deep learning is a specific type of machine learning algorithm that uses multi-level neural networks.

A neural network is an algorithm that simulates the structure of the brain by having nodes (neurons) and relationships (dendrites). A neural network solves a problem by converting the problem into a set of input nodes, a set of output nodes, and a set of intermediate or "hidden" nodes. Deep learning is just a neural network with multiple intermediate layers of nodes.

Machine learning is the subset of artificial intelligence that deals with computers learning to solve problems, and optimize solutions by analyzing data instead of being directly programmed.

There are many different types of machine learning algorithms, including deep learning, generic algorithms, reinforced learning, and many others. Any algorithm that can modify itself and learn from its data can be considered machine learning.

Artificial Intelligence is the field of computer science that deals with algorithms and technology for implementing and simulating complex cognitive tasks.

Artificial Intelligence includes many types of algorithms, including deep learning, genetic algorithms, fuzzy logic, state machines, expert systems, and many more.

You can associate a user question or phrase, or individual words with sentiment (good vs bad). Sentiment can be used to track how the user is feeling about the current conversation. The sentiment statistic is tracked by your bot daily and can be graphed from its Analytics page.

Here is an example response list that defines sentiment for words and phrases:

You can get to the Admin Console for a bot or live chat channel from its webpage.
Click on the menu (hamburger) button or admin (gear) button below the "Chat" button.

how do I find the admin menu

What are required words and how do you use them

The syntax to define a pattern is,

Pattern("Give me * or *")

Hi,

I have encountered an issue on this bit.

I am not able to train my bot for the response using this response: Template("I ll give you {star[0]}") for the question Pattern '('Give me * or * ')'

The response possible comes from a list i guess and in some other examples that I tried, the response is as same as the question.

Please let me know on this. Thank you.

I'm trying to use speech recognition but i'm getting the error: function toggleListen doesn't exist

The issue you where browsing your bot's "conversations" logs and click on the button to "correct" the bot's response. This enters a new response to the "question" that the bot was asked. In this case the "question" is "hockey", so you are entering a "response" to "hockey".

Your issue is you are trying to enter a "question and response", just enter the "response".

Many thanks for the quick reply and the links.

I will investigate accordingly.

with regards,

Dennis

Actually if you look at the post it says, "Updated: Nov 11, 13:14", so was updated just this month.

We have a script library of many different examples here,

This page has not been updated in two years. 

Are there working source files we can see of the above examples? Esp. the csv example, it is not  clear. does the csv need column headers? How are the words separated in the cell itself? 

input	response	keywords	required words	topic

Bot Libre's scripting language Self emulates JavaScript, with a few extra enhancements such as states, patterns, and templates.

Self includes several class for useful operations that you can use in your scripts.

When you embed a bot on a webpage using the "Box" embedding option it will put a chat box on the bottom corner of your page.

The embedding code that you can get from your bot's Embed button contains the HTML/JavaScript code you need to paste into your webpage to embed your bot.

In your bot's Twitter properties you can configure it to retweet or process/reply to tweets with specific keywords or hashtags.

Normally a bot will perform a Twitter search to find new tweets for it to process. You can configure your bot's Twitter search in its "Tweet Search" property.

Bot Libre supports integrating bots with many different communications and social media platforms. Each platform provides a different API that has different functionality.

For some integrations, such as Facebook Messenger, and Telegram, their API's support real-time messaging, so responses are instant, or only a second or two.

Emotions, actions, and poses can used to interface with your bot's avatar.

Avatars are a collection of media files (images, video, audio) that have been tagged with emotions, actions, and poses. You can choose your bot's avatar from its Avatar page in its Admin Console. You can choose any existing avatar from our avatar library, or create your own.

There are many things that can affect your bot's performance, and many things you can do to improve your bot's performance.

Learning Options

When you embed your virtual agent or chat bot on your own website there are several methods to do this. You can get the code to embed your bot from your bot's Embed button on its page.

Your options include:

I tested the same log on the website and it worked correctly as expected.

Your issue is your are importing the file as a "Chat Log" not as a "Response List". The chat log format is for a conversation log, the bot can learn from a chat log in context when learning is enabled. What you want to load is a response list, which is a list of question/response pairs with additional info like keywords, labels, etc.

I entered your chat log exactly as posted.

Hello
Hello, how may I be of service?
label: #hello

Hi
#hello

How is it going?
#hello

Here is my conversation:

Topics can be used to add context to a response, or conversation.

Topics can be set on a response in the Training & Chat Logs page in your bot's , or by importing a response list, or AIML file.

Bot Libre supports labels. A label lets you name a response and then reuse it just using the name. You can set the label for a response from the Training & Chat Logs page in your bot's Admin Console, or by importing a response list file.

To reference a label just put the label name as the response i.e. #hello (labels must start with a #).

The read_page_mailboxes permission is required to be able to respond to messages for a Facebook Page. Currently Facebook has not granted the Bot Libre app this permission, so this functionality only works from our own Facebook pages right now.

We are still hopeful that Facebook may grant us this permission, but they seem to be very restrictive on their API usage.

Hello, 

my bot is having this log error: 

WARNING -- FacebookMessaging:message - (#279) Requires extended permission: read_page_mailboxes code - 279 Relevant information for error recovery can be found on the Facebook Developers Document:

One of the issues is that you have added a bunch of scripts to your bot. In particular the "SimpleLanguage" example Self script. This script attempts to understand questions and gives answers like,

"I understand, that might religious"

I enabled "Response matching" but it's still forcing me to teach it every possible combination for the phrase "are you religious" (which has lots of prefixes and add-ons) without using keywords or any tool it has to come up with responses that make sense on its own.

Twitter bots will automatically respond to mentions and messages every time they check their status. How often the bots check their status depends on your account and server.

On www.botlibre.com, basic accounts will only check their status once per day. You can upgrade your account to Gold or Platinum to check every hour, or 20 minutes.

Is there any way to make it so that I don't need to click "Check Status" every time so it can respond to a person? I want it to automatically respond to a person after they mention them instead of having to click "Check Status" every single time.

You can manually test your bot using the "Check Status" button from the Twitter page from your bot's "Admin Console".

First go to the "Log" page and set logging to "FINE" then click on "Check Status" from the "Twitter" page, and go back to the "Log" page to see what occurred.

Bot Libre support speech recognition on both web and mobile.

Web

Looks like you imported the AIML as a chat log. You can either import AIML as a chat log, from the Chat Logs page, or as a script, from the Scripts page.

Bot Libre has multiple ways to resolve responses. It will first look for a direct response match, then process its program scripts, they search for the closest matching know response.

The content of the AIML files appears on the page, but it is not added to the list of scripts.