Sentiment, Emotion, Attitude, and Personality, Via Natural Language Processing

(I wrote this article for the IBM Watson blog, where it first appeared.)

It’s a privilege to have Rama Akkiraju, IBM distinguished engineer and master inventor, participate as a Vision and Opportunity panelist at the 2016 Sentiment Analysis Symposium. I organize the symposium – this year’s event takes place July 12 in New York – and recognize the many ways IBM has, over the years, expanded what’s possible in the realm of what I’d characterize as “human data.” (Disclosure: IBM is a symposium sponsor.)

RamaAkkiraju
IBM Distinguished Engineer Rama Akkiraju

What’s Rama up to, at IBM?

“My team at IBM has been focused on developing technology to better understand people at a deeper level based on sentiment, emotion, attitude, and personality,” said Rama. “With our work with Watson APIs – such as Tone Analyzer, Personality Insights, Emotion Analysis, and Sentiment Analysis – we’re working to enable more compassion, engagement, and personalization in conversations across various channels.”

Sixty Years of Natural Language Processing

The core text-analysis technologies – applying statistical and linguistic methods and machine learning – aren’t completely new. IBM’s Marie Wallace, a 2014 sentiment symposium speaker, relates in a blog article that she “joined IBM in 2001 to build the next generation of NLP technology for IBM… the 3rd generation of IBM LanguageWare, which initially started back in the ’80s.” And I wrote, myself, in a 2008 InformationWeek article, BI at 50 Turns Back to the Future, about 1950s work by IBM researcher Hans Peter Luhn on the creation of business intelligence via text analysis. But Watson embodies a qualitative leap ahead, including in its advance into personas and emotion analytics.

According to Rama, “using the Watson APIs to detect emotions and tones in a conversation can be very powerful in many contexts such as customer service and health assistance telephone hot lines.” She explains that “at this juncture, the Emotion Analysis API detects five emotions – Joy, Fear, Sadness, Disgust and Anger. We’re currently working on expanding more positive emotions for Watson to detect. Given the broad spectrum of emotions humans experience by associating with different entities, we want to expand emotion detection to inform and shape more compassionate responses in a conversation.”

Watson Tone Analyzer output
Tone Analyzer output (https://tone-analyzer-demo.mybluemix.net/)

IBM Watson avatarThe Watson Developer Community: InMoment and Pulsar

Watson is a cognitive system made available via the Watson Developer Cloud, a collection of analytical services. A point of pride for IBM: “Watson is open to the world, allowing a growing community of over one million developers, students, entrepreneurs and tech enthusiasts to easily tap into the most advanced and diverse cognitive computing platform available today. Hundreds of clients and partners across six continents and across dozens of industries actively use Watson. More than 550 are already commercializing their ideas and over 100 of these partners have already introduced commercial cognitive enabled apps, products and services to the market.” Several of my consulting clients work with the technology.

Customer experience solution provider InMoment’s co-founder Kurtis Williams spoke to me about his company’s IBM partnership. Kurt explained, “we use two different pieces of Watson text analytics: the Watson Explorer (WEX) and the Watson Explorer Content Analytics Studio (formerly known as LanguageWare). WEX is a text mining tool that we use during professional services engagements and is very useful out of the box and handles large amounts of text elegantly. It is based on Watson’s linguistic technology coupled with search engines and a faceted search exploration interface.” The value for this particular partner? “We consider this library of IP to be very important to our overall text analytics strategy.” Talk to Kurt about InMoment’s work with Watson; he’ll be attending the 2016 symposium.

Kurt spoke at last year’s conference, as did another Watson partner, Francesco D’Orazio, product and research VP for social data intelligence platform Pulsar. According to Fran, “Pulsar integrates Watson as part of Modules, our Artificial Intelligence on Demand solution that allows users to deploy various services to analyse their data, depending on their use case. We currently integrate the Concept Tagging for Images, Text Extraction from Images, Emotion Analysis, Personality Insights and we’re planning on integrating more services by the end of Q3.” Fran’s 2015 talk was titled “Analysing Images in Social Media. Fran explains, “we chose to work with Watson amongst our top-tier partners because of their emphasis on cognitive computing and the fact that a single integration gives our client access to a wide variety of machine learning and deep learning tools supporting many different use cases.”

Vision and Opportunity

Looking ahead – and that’s what Rama and fellow Vision and Opportunity panelists at the Sentiment Analysis Symposium will be doing –Rama relates, “we are moving toward an era of computing that is based on cognitive systems like Watson that learn, reason, and adapt. These systems are increasingly focused on making technology more natural for humans to interact with in intuitive ways such as conversations in natural language.”

Rama continues, “as a result of this ongoing shift, my team is working to develop systems that can better gauge users’ specific needs and individual contexts at a given time to provide the most relevant information or service. More effective communication in the future will be shaped by Watson helping people detect tones, emotions, and sentiments from different forms of human expression including text, voice, gestures, and facial expressions.”

Computing that understands and synthesizes the full range of available data – both fact and feelings – to meet individual users’ situational needs is ambitious but achievable. It is the day to realizing a world of opportunity.

Leave a Reply