AI News, The Microsoft India Blog The Microsoft India Blog
- On Monday, June 4, 2018
- By Read More
The Microsoft India Blog The Microsoft India Blog
Detecting emotions in text is difficult enough for human beings, let alone artificially created machines, as many of our emotions are conveyed through expressions and tone of voice.
Here’s a look at how we arrived at a novel approach to detect emotions in textual conversations: The challenge of creating emotional Artificial Intelligence (AI) Detecting emotions accurately is always a challenge, even for human beings.
Although these gestures are subtle and easy to miss, they carry a wealth of information that can add context to the conversation.
A machine that can detect emotions can generate responses that genuinely help users seeking assistance or information.
Training wheels for AI High-quality and high-volume data combined with appropriate labels is essential when solving a machine learning problem like this.
The human judges used textual cues, smileys, emoticons, and punctuation as clues to detect emotions throughout the set.
For example, if a person responds to a text with “there, there,” it could convey a need to comfort someone which may indicate a sadness or despair in the previous text.
Once the human judges pruned this relatively small dataset, a nearest neighbour based clustering algorithm was used to automatically sort the larger set into respective categories.
As the name suggests, this Deep Learning model combines semantics and sentiment indicators to categorize text conversations based on the emotions they convey.
To train the model, we used the Microsoft Cognitive Toolkit and divided the data into ‘training’ and ‘validation’, based on a ratio of 9:1 (9 sets for training and 1 set for validation).
The results For each of the four emotion classes, our SS-LSTM model outperformed all other known state of art techniques for detecting emotions in textual conversations.
By creating a model that combines semantics and sentiment indicators in short, text-based conversations, our team has helped take a significant leap forward in detecting emotions.
- On Wednesday, January 16, 2019
How to Have a Good Conversation
We too often imagine that 'good conversations' are things we fall into out of luck. Far from it, knowing how to have a good conversation is a skill that can be learnt ...
Natural Language Processing With Python and NLTK p.1 Tokenizing words and Sentences
Natural Language Processing is the task we give computers to read and understand (process) written text (natural language). By far, the most popular toolkit or ...
Science Of Persuasion
This animated video describes the six universal Principles of Persuasion that have been scientifically proven to make you most ..
Tom Snyder and Satish Jha in conversation about Emotion Mining, Motivational Intelligence, Emotional Intelligence and how to engage people for branding and ...
Seeing Behaviors as Humans Do׃ Uncovering Hidden Patterns in Time Series Data w⁄ Deep Networks
Time-series (longitudinal) data occurs in nearly every aspect of our lives; including customer activity on a website, financial transactions, sensor/IoT data.
Brain Tricks - This Is How Your Brain Works
Get the book: TWEET VIDEO - Ever wonder how your brain processes information? These brain tricks and .
AI vs. AI. Two chatbots talking to each other
Are you a Robot or a Unicorn? Let the world know: What happens when you let two bots have a conversation? We certainly ..
Charlie Chaplin - Final Speech from The Great Dictator
Subscribe to our channel: • Get it on iTunes: • "We think too much and feel too little. More than .
How to control someone else's arm with your brain | Greg Gage
Greg Gage is on a mission to make brain science accessible to all. In this fun, kind of creepy demo, the neuroscientist and TED Senior Fellow uses a simple, ...