Teach on Udemy

Turn what you know into an opportunity and reach millions around the world.

Learn More

Your cart is empty.

Keep shopping

Sentiment analysis for chatbots - DialogFlow, IBM Watson

Name: Sentiment analysis for chatbots - DialogFlow, IBM Watson
Rating: 4.5 (30 reviews)

Sentiment analysis for a chatbot with Google DialogFlow, IBM Watson, AFINN. Sensitive chatbot with sentiment analysis

Created byJana Bergant

Last updated 5/2020

English

English [Auto],

What you'll learn

How sentiment analysis can benefit a chatbot
Common implementations scenarios for a chatbot
Use DialogFlow's sentiment analysis to make chatbot emotionally sensitive. You'll have a chatbot that can read the sentiment of the user's message and act on it
Learn how to use IBM Watson for sentiment analysis
Learn how to use AFINN sentiment analysis

Course content

6 sections • 34 lectures • 2h 41m total length

Introduction4:21
Explore how to implement sentiment analysis for chatbots, using dictionary-based and machine learning approaches with Google natural language and IBM Watson, to detect frustrated users and trigger Slack alerts.
Tech-stack3:45
Explore the tech stack for a sentiment-aware chatbot, including a Node.js and Express backend hosted on Heroku and Google Dialogflow for natural language processing with sentiment analysis.
Behind the scenes - bot architecture2:35
Learn how two bot apps connect via Facebook Messenger, a Facebook page and app, a backend, and Dialogflow to analyze messages, detect intents, trigger actions, and generate responses.
How to get help1:13
Additional resources0:25

Sentiment analysis in a nutshell5:36
Understand the emotional tone and opinions in text with sentiment analysis. Analyze unstructured data at scale to deliver real-time insights and consistent business decisions.
Sentiment analysis for a chatbot5:35
Explore real-time sentiment analysis for chatbots, identifying key words and assigning positive, neutral, or negative values to guide responses with context and tone.
Common implementation scenarios in chatbot3:04
Explore common sentiment analysis scenarios for chatbots, detecting positive, neutral, or negative sentiment to tailor dialogues. Set thresholds for handoffs, score customers, and identify brand ambassadors for marketing campaigns.
Sentiment analysis examples - AFINN sentiment lexicon6:41
Explore dictionary-based sentiment analysis with the AFINN lexicon, scoring tokens and computing a comparative score, demonstrated via a Node.js sentiment module and negation handling.
Sentiment analysis examples - Google NLU9:07
Learn how to use Google's natural language understanding API to analyze sentiment and entities, interpret sentiment scores and magnitudes, with a practical demo and authentication setup.
Sentiment analysis examples - IBM watson9:19
Learn how IBM Watson sentiment analysis works with emotion scores and entity sentiment, using API keys, curl and code examples to analyze text like 'I love apples'.
How does sentiment analysis work? - Part 13:20
Explore how sentiment analysis works for chatbots by comparing rule-based, machine learning, and hybrid approaches, including negation handling and simple positive/negative word counts.
How does sentiment analysis work? - Part 23:22
Learn about automatic sentiment analysis using machine learning to classify text as positive, negative, or neutral. Explore training and prediction, feature extraction, and vector representations from lexicons to word embeddings.
How does sentiment analysis work? - Part 34:58
Discover how sentiment analysis uses naive bayes, logistic regression, SVMs, and neural networks with feature extraction, evaluated by cross-validation, accuracy, and F1 in a hybrid rule-based approach.
Sentiment Analysis tools & APIs2:19
Explore ready-made sentiment analysis APIs, including Google Cloud Natural Language, IBM Watson, Amazon Comprehend, MonkeyLearn, Rosette, and SAS API, to avoid building from scratch and scale across languages.

My bot with AFINN sentiment analysis1:04
Apply afinn sentiment analysis to a personal bot, map text messages to sentiment before responses, and review the code to prepare for implementing Google sentiment analysis.
Me bot code overview11:04
Demonstrates routing messages from a Facebook page through a Node.js Express app to dialog flow for sentiment-aware intent, actions, and live agent handoff with Slack alerts.
Get sentiment result of user message3:14
Learn to implement sentiment analysis in a chatbot by installing and integrating a sentiment module, analyzing user messages to produce a score, and routing low sentiment to a life agent.
Create sentiment service module4:13
Create a separate sentiment service module, move sentiment analysis into it, and store per-user sentiment history in a time-stamped map with add and get methods for score and comparative.
Route to LIVE agent when sentiment is too negative4:03
Analyze user messages with a sentiment module, store scores by timestamp, and route to a live agent when the latest sentiment falls below -2, ensuring proactive human handoff.
Intervene when sentiment changes4:37
Learn to detect a sudden negative shift in user sentiment by comparing current and previous scores, and intervene with empathy by routing to a life agent when thresholds are crossed.
Slack webhook verification8:39
Learn to verify a Slack webhook, configure a bot app with authentication tokens, and subscribe to direct messages to route and post messages to the correct channel.
Send notification to slack when live agent needed6:29
Learn to trigger Slack notifications for live agent needs by reading bot events, extracting the channel, and posting messages via the Slack chat.postMessage API.

How to get Smartbabe working - FB page, FB app8:40
Learn to get the smart babe chatbot running on Facebook Messenger by creating a Facebook app and page, connecting to Dialogflow, and enabling basic sentiment analysis in the workflow.
How to get Smartbabe working - Dialogflow8:24
Set up a Dialogflow agent for Smartbabe, create a Google project and service account, import intents and entities, and configure training phrases and responses for sentiment-enabled chatbot.
How to get Smartbabe working - backend app, webhook validation10:17
Build a sentiment-analysis chatbot backend by wiring a Facebook app, Dialogflow, and Heroku, validate webhooks, and configure environment variables and Google service accounts for deployment.
Code overview2:46
Explore integrating the Google natural language API into a chat bot via DialogFlow, focusing on extracting responses and sentiment analysis, handling flow responses, and upgrading DialogFlow for sentiment support.
Enable sentiment analysis in DialogFlow2:24
Enable sentiment analysis in DialogFlow by upgrading to enterprise edition, configure Google Cloud Natural Language, and test a sentiment score of 0.8 with magnitude 0.8.
Check sentiment analysis results1:44
Enable sentiment analysis in the dialogue flow, access sentiment scores in the response function, and understand that zero or neutral sentiment may not return a result, with module upgrades required.
Call live agent on low sentiment3:12
Enable sentiment analysis in dialog flow, log user sentiment history with score and magnitude, merge new results with existing data, and route low sentiment users to a live agent.
Intervene on sentiment change4:54
Monitor sentiment history in DialogFlow and IBM Watson chatbots and intervene when sentiment shifts too much, using thresholds like 0.5 and -0.6 to escalate to a live agent.
Sentiment analysis challenges today4:39

Requirements

Knowledge of nodejs programming language
Knowledge of GIT

Description

In this course, you'll learn how to implement a sentiment analysis to a chatbot. You will teach your bot how to be emotionally sensitive. And most urgently how to spot a frustrated user and save the day.

In the course, we'll be working with my code from the messenger chatbot course. Even if you did not take that course, you'd be able to take the code and work with it.

We'll use Node.js for programming and GIT for deploying and version control. The bot will is hosted on Heroku, but you can simply host it anywhere else where they support Node.js. We'll use DialogFlow to process natural language. DialogFlow will help us understand what users want.

In the process, you'll learn how sentiment analysis works behind the scenes and how to implement sentiment analysis to a chatbot.

In the first section, we'll take an overview of the app infrastructure and get familiar with the tech stack, which is the technology used in this course.

The second section will introduce you to Sentiment analysis, what is it and how it can benefit a chatbot. We'll look at common implementation scenarios in a chatbot.

Do you want to know the difference between the rule-based approach and automatic approach with machine learning? We'll go through examples of both and make a little demo for each. You'll get familiar with dictionary-based solution AFFIN and Google's natural language API and IBM Watson natural language solution. I will gently introduce you to the algorithms they use and how they work. So you know how things work inside that black box.

Then in the third section, we start implementing a dictionary-based sentiment analysis. We'll use my chatbot as a sample. First, I'll give you an overview of the code and show you where and how to get a sentiment result for the user message. Then we'll create a sentiment analysis module and teach a bot how to route to live agent upon identifying a super frustrated user. We'll also teach a chatbot to be sensitive to sentiment change and intervene when it detects user's experience satisfaction is decreasing. And then I'll give you another candy. I'll show you how to send a notification to Slack when a frustrated user comes along.

When you know how to implement a dictionary-based solution, we'll move to a machine learning solution. And that is what we'll do in section 4. One of the best is Google's natural language solution. We will do the full implementation into a chatbot.

And in this demo, we'll use a chatbot from my previous course. You'll get the source code and will be able to work with it even if you did not take that course.

After the course, you'll have a demo bot with built-in sentiment analysis and knowledge of how to implement it into your chatbot.

To make it easier for you I've added git commits with changes to every video that has a change in the code. This way, you can compare your code to mine. And you won't lose time debugging. And remember for any questions I'm available in the Questions and answers. You are not alone.

And remember, I'LL BE THERE FOR YOU.

My name is Jana, and I'll be your instructor in this course. I'm a web engineer with 20 years of programming experience. I'm also an IT instructor teaching people new tech skills. Over 17000 people are already taking my course.

I help all my students at every step of development. And I'll be here for you!

At the end of the course, you'll have a chatbot that can read the sentiment of the user's message and act on it. Wait no longer. Take the course and make your chatbot even better.

See you in the course!

Jana

Who this course is for:

chatbot owners that want to improve chatbot performance

Sentiment analysis for chatbots - DialogFlow, IBM Watson

What you'll learn

Explore related topics

Course content

Introduction5 lectures • 12min

About Sentiment analysis10 lectures • 53min

Implementation for a chatbot - 1 example8 lectures • 43min

Implementation for a chatbot - 2 example9 lectures • 47min

Platforms, NLP & Libraries for Voice Bots & Chatbots1 lecture • 5min

Conclusion1 lecture • 1min

Requirements

Description

Who this course is for: