Online influencers are powerful.
These thought leaders, emerging businesses, and trending personalities have an impact on business like the world has never seen before. Over 70 Percent of Instagram users have made purchase decisions after seeing influencers’ postings (source).
The ability to easily identify these powerful influencers and analyze what they say can change the game when it comes to industry analysis and decision-making.
Wissee Influencer Discovery and Expertise Tagging System makes that possible.
It’s the first deep learning model trained on large-scale and heterogenous industry data. What that means is that we use millions of social media postings, company reports, Wikipedia pages, and patents to develop our proprietary AI techniques, which can serve customers with the high-quality information they need to know.
Here’s what makes this system valuable
Our influencer discovery process includes:
1. Username recognition
2. Network analysis
With username recognition, Wissee extracts entity names from a brand or influencer’s social media postings and then analyzes their interaction networks, so we can dig in to find and expand the automatic influencer discovery.
Here’s an example of Influencer Discovery in action:
Instagram is currently the most popular social media platform for influencer marketing. However, there is a technical barrier for extracting user account names from text.
Here are examples of IG posts collected from Wissee’s API service.
Examples of Instagram Postings from API
The detected usernames are underlined.
“we’d like to take a moment to respect this flick babe alissajanay used anastasiabeverlyhills Liquid Liner to create this look - link in bio to shop!”
“Naruto fan Jlin7 wearing an Adidas Pro Bounce 2018 Low player exclusive against Sacramento.”
The traditional entity recognition tool or word embedding technology cannot recognize social media usernames. Wissee can.
Our trained, character-level based AI model makes it possible to recognize usernames across over one million social media postings. Our model is the first in the market and shows 82% accuracy, too.
Once an influencer has been identified, analyzing his, her, or their expertise is important not only to quantify the quality and efficiency of brands' influencer marketing but also to discover more and better influencers.
To do that, Wissee offers entity embeddings that help to extract important influencer information. Entity embedding sets up a foundation for various applications, such as measuring the influencer’s similarity to a brand and expertise tagging.
Our feature extractor uses millions of heterogeneous industry data points collected and labeled by Wissee’s proprietary techniques.