How Atomic AI Measures The Emotion In Your Content

How Atomic #AI Measures The Emotion In Your #Content - by @suhash_talwar @atomic_reach

  • Why Writing With Emotion is Important
    Emotional language can greatly impact how engaged a reader is with your content.
  • The emotional factor was looked at on a per article basis – using “hot” and “cold” as the indicator for the overall emotional intensity of a piece of content.
  • Soon after, we updated the feature to flag specific words as hot (emotional) or cold (not emotional) so that users had a better idea of what words were contributing to the level of emotion within a piece of content.
  • We noticed that our users would look at the emotional intensity of a piece of content, then haphazardly try and replace words.
  • After various iterations, the team was finally able to settle on a solid model for measuring emotion and was able to bake in the ability to provide recommendations to either increase or decrease the emotional intensity of the word.

If there is a single feature in our platform that has generated the most interest (but also the most confusion), it is our Emotion measure.

@Atomic_Reach: How Atomic #AI Measures The Emotion In Your #Content – by @suhash_talwar @atomic_reach

If there is a single feature in our platform that has generated the most interest (but also the most confusion), it is our Emotion measure.

Clients would be intrigued by the concept of intelligently balancing how much emotion should be in a piece of content. Each and every time the emotion measure would elicit the most “oohs” and “aahs” during our demos.

But when it came to actually using it within the platform, clients would continuously stumble over this feature. In the beginning, unlike the other measures, the logic behind the ’emotion measure’ was sort of flawed. The UI of  Atomic AI didn’t do a great job in giving the user a clear sense of “what to do” once it was activated.

To really understand what the Emotion Measure is all about and how it can help you, we interviewed key architects Paul, Emilian and Jessica. Here’s how they revamped and improved Emotion in Atomic AI.

Emotional language can greatly impact how engaged a reader is with your content.

Viral content is a good example of this since it is written in a way that ramps up the emotional intensity. It elicits a reaction or feeling in the reader that compels them to talk about it or share it.

Recent coverage on the impact of “fake news” further solidifies that highly emotional content will get your audience to respond. Regardless of how preposterous the claims are, if written in a highly emotional manner, and delivered to the right audience, it can elicit an emotional response.

On a less dour note, introducing emotion into a piece of business content can do wonders for engagement. This is especially true for B2B, which sometimes has the tendency to be dry and rather uninspired. There is an ongoing debate about the need for emotion in B2B content, with many rejecting the idea while others embracing it.

Bennet Bayer, CMO at Huawei Technologies states that “many buyers can see this as manipulation and resent it” in regards to emotional language in B2B content. In contrast, Nicc Lewis, VP Marketing at Leverate points out that:

The key takeaway here is balance – too much emotion and your content starts sounding like a Buzzfeed article (not good). Not enough emotion will result in a dry, uninspired piece of content.

When the emotion feature was first introduced, it was little more than a status indicator. The emotional factor was looked at on a per article basis – using “hot” and “cold” as the indicator for the overall emotional intensity of a piece of content. The user would have no idea about what words to change or what to replace them with to increase or lower the overall emotional intensity of the article.

Soon after, we updated the feature to flag specific words as hot (emotional) or cold (not emotional) so that users had a better idea of what words were contributing to the level of emotion within a piece of content.

We noticed that our users would look at the emotional intensity of a piece of content, then haphazardly try and replace words. This resulted in a lot of trial and error and frustration for our users as they lacked a proper way of making measured decisions in regards to emotion in their content.

Our users continually asked us to better explain this feature and provided a lot of feedback on how to make it more valuable and user-friendly.

During the testing period, we broke down emotional words into 3 components; dominance (red), arousal (green) and valence (yellow). The problem here was that when you replaced a red word, it would in turn create a green word. It was a downward spiral in a sense, where fixing one problem would create a different one.

This proved to be unwieldy, and lead to focusing on just a single measure of emotion. With this improvement, we still had the same problem of “what now?” – users had no idea how to actually make use of this measure.

We knew that we wanted to provide recommended synonyms, much like our readability measure. This meant that if a word was flagged as hot or cold, the user would have the option to either increase or decrease the emotional intensity of the word. After tons of research and a few weeks of planning and implementation, the platform was now able to give you a list of recommended synonyms to improve your content emotionally via the click of a mouse.

After various iterations, the team was finally able to settle on a solid model for measuring emotion and was able to bake in the ability to provide recommendations to either increase or decrease the emotional intensity of the word. This enabled the user to not only understand the intensity of a particular word, but choose whether or not they wanted to increase or decrease the level of emotion.

It was the missing piece in the emotion puzzle and also created a sense of uniformity with all the other measures in the platform.

Once the team had perfected the synonym recommendations for emotional intensity, it introduced another problem. Adjusting the emotional intensity of a word by choosing a different synonym often changed the readability of that word or sentence. This would then clash when the readability measure was turned on – resulting in a never ending loop where one edit would affect the other measure.

The team went back to the drawing board and came up with an intuitive solution where the user could now choose a high or low intensity synonym that maintained the same readability level. This meant that every recommended synonym provided would be at the same readability level. This was an elegant way to remove the aforementioned loop of one edit affecting the other. It also maintained consistency with the rest of the platform by focusing on readability of the content.

What’s Coming Up Next?

Being a company that works under an agile model, iterations are constant and we are always evolving the product. Through feedback from the rest of the team and current clients, there are many improvements that are planned for the Emotion Measure.

The main focus now is on developing a more accurate recommendation within this measure. Being able to provide synonyms that would make the more sense in the context of the sentence would have an added benefit of increasing the accuracy of synonyms recommended under the readability measure.

With 2017 trends foreshadowing data as the driver for marketing success, where do you see sentiment and user-behaviour in the flow of your content marketing strategy?

How Atomic AI Measures The Emotion In Your Content