Once chosen, the user can select to cluster objects by:
Social Relatedness: Takes into account the explicit and implicit meaning of an object and the different ways in which users talk about it on social media
Demographics: Takes into account which demographics are most commonly talking about an object
Psychographics: Takes into account the general interests of those talking about an object
Geography & Language: Takes into account the location and language of the person mentioning the object
Monthly Activity: Takes into account an object’s monthly post volume
Penetration: Takes into account an object’s average percentage of the focus subject’s conversations
Correlation: Takes into account the explicit meaning of an object and how it is talked about on social media
Based on which factor is selected, objects are grouped together based on similarity within the chosen factor. For example, correlation is selected in the image below, and objects ‘oiliness’ and ‘oily skin’ are clustered together, meaning they are used almost interchangeably in conversations on Instagram. ‘Frown lines’ is the farthest object away from ‘oiliness,’ meaning these two objects do not have interchangeable explicit meanings based on how they are referenced on Instagram.
The Multi-Trends view looks at each object’s change in a variety of metrics over the past 1, 3, 6, and 12 months (image 6). Objects can be sorted alphabetically, by metric, or by the largest change over any of these time intervals by clicking on the desired option from the bar highlighted in blue in the image below. |