Bye Booleans - Hi Talkwalker AI Engine. How does AI work in social media analytics?
How does AI work in social media analytics? Talkwalker is leading the way as a pioneer in artificial intelligence. Here’s what we’re doing. With the Talkwalker AI Engine combining our image recognition, and sentiment analysis, plus our revolutionary new custom models, you now have access to more high quality data and digital marketing insights. Even faster.
The Talkwalker AI Engine - Less time analyzing, more time acting
The Talkwalker AI Engine puts PR, marketing and advertising pros back in control. With extra time, you can focus more on using your data to find useful business insights.
- You can boost your content creation, knowing exactly how your audience is engaging.
- You can maximize your impact by targeting audiences more relevant to your brand, with engaging content marketing.
- You can improve your customer experience, by monitoring every conversation about your brand (and the sentiment behind them).
The possible AI marketing applications for the Talkwalker AI Engine are almost limitless, and will drive impact, and financial results for your brand, for years to come...
The death of Booleans
But they have their limits -
- You need an understanding of social listening to write them effectively.
- You risk missing data if your Boolean is too precise.
- You can get too many false positives if your Boolean is too broad.
- You can spend several days adjusting and testing your Boolean, to ensure you’re covering the precise data you need.
Our AI-powered custom models remove the need for complicated adjustments, give more accurate results, and will be up and running in around 30 minutes.
Yes. You read that right.
Our most accurate data sets in just 30 minutes! And without the technical know-how. I wasn’t kidding when I said this was revolutionary.
Let’s take a look…
Our brand new AI-powered custom models have arrived
Let’s look at Apple. One of the biggest brands in the world, with over 102 million mentions in the last 12 months. If you were monitoring this as your brand (or competitor’s), it would be extremely complicated to create an accurate Boolean for it.
There are numerous contexts people could use to discuss the brand. Apple iPhones, Apple Music, Apple TV, Apple MacBooks, Apple HomePods. The list goes on.
And the list of false positives is long too. Apple could be mentioned as a fruit, a flower, a tree, a color, an ingredient, a film, a band, an album, a town in Switzerland, a river in Illinois, etc., etc., etc.
People may even mention it in proverbs. But probably only the bad apples.
To clean that sort of data, it used to take a lot of hard work.
The old Boolean way
To monitor that range of mentions, while minimizing the number of false positives, takes a lot of trial and effort.
With a simple query, we get 2 million results in 7 days. But with an accuracy of just 90%. (That may sound good, but that equates to around 200,000 bits of irrelevant data.)
With experience, you can create a Boolean to narrow your results. This takes time, as you have to constantly reassess the quality of the data, then consistently readjust the Boolean to maximize your data accuracy. Each improvement will create a more and more complicated Boolean, so you end with something like this…
Even then, we can’t be certain we’re collecting all the relevant data we need. Or that our project is 100% clean. For this example, we collected 1.7 million results over 7 days. So yes, we improved the accuracy to 93%, but lost some relevant mentions along the way.
The new Talkwalker AI-powered way
This is where the Talkwalker AI Engine comes to the forefront.
Talkwalker is the pioneer of artificial intelligence and social analytics. The new custom models use natural language processing, vast computing power, and deep learning, to interpret large amounts of data quickly.
With minimum setup, you can use the power of AI and machine learning, to automate data segmentation. And it’s ridiculously simple to do...
1. Create your query. But keep it simple. In this case, all we need to search for is mentions of Apple.
2. Start-up the Talkwalker AI Engine in the Project Settings.
3. Decide your categories. All we need for this is one for Apple, the brand. The rest will be irrelevant.
4. Then get classifying. The AI Engine will show you mentions relating to your query. All you have to do is decide whether they relate to the Apple brand...
5. Or not…
6. Just keep clicking. As you classify more and more mentions, the progress bars will fill up. Once full, just click train.
7. The Talkwalker AI Engine trains in a matter of seconds, and then applies your classifications across your entire data in real-time. You can apply your models to back data too!
8. Use all the time you’ve just saved to look for insights that are relevant to you.
In this example, it took only 30 minutes to classify the relevant mentions of Apple in the AI Engine!
That’s a huge amount of time saved on the previous Boolean method. And provides us 1.8 million results (an extra 100 thousand) with a phenomenal accuracy of 99.5%.
Talkwalker AI Engine to boost brand impact
The new AI-powered custom model is at the forefront of our AI research, and opens up a variety of analytical options. You can create numerous categories to classify your data with, helping you solve a variety of problems, like...
Powerful classification of all your brand or product features - no matter how complicated
What if you want to accurately monitor a specific aspect of your product? Or just one of the brands your business has to offer?
And what do you do if your consumers describe your brand in numerous ways?
Marina de Tavira has had a long theater career and is soon to be seen in Alfonso Cuarón’s “Roma.” She has recently appeared in the series “Ingobernable” for Netflix and “Falco” for Amazon https://t.co/NibFXKCIGf pic.twitter.com/UtQyW77W32— Variety (@Variety) August 21, 2018
How do you monitor mentions of Amazon Prime Videos, when people often call it just Amazon?
With the Talkwalker AI Engine, you can classify numerous products or features for easy comparison.
For example, Amazon is a global brand with hundreds of aspects to their business. Monitoring Amazon would only give you a broad overview of the brand. But by adding classifications like Amazon Prime Video or Amazon Kindle, Amazon could monitor how each element of their business is perceived.
This method beats Booleans, as you often find posts with multiple brand product mentions. You can train the AI Engine to classify the product(s) you want to prioritize, to give you a more focused brand picture.
Using the Talkwalker AI Engine to classify specific Amazon products, it’s clear Amazon Prime Video is more popular with 3 times the mentions of Amazon Kindle. Marketers now have clearer insights into which products are more, or less, popular, and can use them to plan marketing strategies effectively.
Track the issues impacting your brand
You can also use the Talkwalker AI Engine to focus on specific issues. Considering Amazon again, they faced two potential crises during Amazon Prime Day this year.
- Technical issues on the website.
- A workers’ strike in Europe.
You can train the AI Engine to categorize Amazon Prime Day mentions, so that Amazon could track the issues. The agility of the engine means this bespoke issue tracking can be set-up quickly, allowing the team to monitor the issue in real time. This helps you plan your crisis response effectively, prioritizing the issues that are more damaging to your brand first.
One keyword. Two issues. Both easy to monitor and categorize with the Talkwalker AI Engine.
Customize your sentiment analysis - to truly know what your consumers love or hate
Talkwalker’s sentiment analysis is already based on the latest AI advances available on the market, providing impressive averages of 90% accuracy. But there are some circumstances where a personally trained sentiment analysis would benefit your brand.
Like if you regularly discuss negative topics to raise awareness.
Sentiment analysis can be trained to suit your project needs. Instead of categorizing mentions by category, you can train the Talkwalker AI Engine with the sentiment of the post. Helping you track emotionally charged topics more accurately.
And, if you’ve already modified sentiment in your project in the past, this will automatically be taken into account when creating a sentiment model. Saving you even more time.
Talkwalker regularly covers issues such as crisis management. Our standard sentiment analysis could flag mentions as negative, but we can train the Talkwalker AI Engine to know otherwise.
And much more
This isn’t the end. The Talkwalker AI Engine will continue to develop, with the ultimate goal of helping you create quality data projects with minimal effort. So you spend less time analyzing, and more time focused on the actionable insights that matter to your brand.
So what’s stopping you? This isn’t science fiction. This is the future and it’s available now. Give it a test drive below.