Okay everyone, now that we've looked at centrality measures using the Social Media Macroscope to understand influence, we'll now jump into another topic with regards to influence specifically surrounding this concept of brand personality. Now, what is brand personality? Brand personality as defined by Jennifer Aaker in 97, is the set of human characteristics associated with a brand. Now she suggested that much like human personality, which is measured in five factors, I'm sure you've heard of openness, conscientiousness, extroversion, agreeableness, neuroticism, she suggested that brands also have a personality, but they have different dimensions. The dimensions are sincerity, excitement, competence, sophistication, and ruggedness. For example, there's just some adjectives that we could look at under each one of these. This is not comprehensive but just exemplary. For example, with sincerity, you have adjectives like domestic, honest, genuine, cheerful. Brands have different ratings according to these different dimensions, how sincere they are, how exciting they are, so on and so forth. Now, a group of researchers and myself, we did a study where we did not use brand personality but human personality and we used machine-learned models that calculate human personality via Twitter feeds. You can read about that in the paper that we provide in this course. But what we found was when we looked at human personality, machine-learned measures via people's Twitter feeds, and then we calculated human personality of the brands that they followed, we found that people tend to follow brands that are closer in human personality congruent to themselves versus brands that are not. We thought that this could be a very interesting method for brands to find influencers that are similar in personality to themselves and they could potentially project marketing that would affect customer satisfaction because their brand personality is much more similar to the brand personality of those brands. Now in this slide, this is just one of the diagrams from our paper because what we were trying to do is we were trying to create not a human personality model, a machine-learned model, but we wanted to create a brand personality, machine-learned model. This diagram is basically showing that we used advanced machine learning Natural Language Processing techniques to build a model that calculates brand personality from social media data. A prototype of this model is accessible via the Social Media Macroscope of which we will use for this course. Basically how it would work is something like this, that we have brand X, and let's say that we have measured their brand personality. Sincerity is one, and this is from a scale from 0-1, their excitement is 0.25, their competence is 0.75, their sophistication is one, and their ruggedness is 0.25. Then we have a blue individual. That person's individual has sincerity one, excitement 0.25, competence 0.75, so on and so forth. Then we have a red individual whose sincerity is 0.25, excitement is one, competence is zero, so on and so forth. If we do a mathematical comparison of these essentially vectors, these five value vectors of brand personality, and you can use a mathematical function, typically called cosine similarity. You can use other similarity measures, but you try to figure out how similar is blue to brand X compared to how similar is red to brand X. What we find here, this is just a contrived example so that we have a perfect match, is that blue is a perfect match in brand personality to brand X, whereas red is not a perfect match to brand X in brand personality. The thought could be that maybe we should engage with this blue individual to be an influencer because they so represent our brand with regards to brand personality. Now just as many of you know that many times influencers on social media, they're not always celebrities that everybody knows. Sometimes they are everyday people just like you and me. But for whatever reason, they built a great following on social media. Maybe they could represent our brand. But how do we know if they're fitting for our brand? How do we know if they could be a good fit to represent what we represent? This brand personality machine-learned modeling might be one way to look at that. In the next lesson, we'll actually dive into looking at calculating this in the Social Media Macroscope.
