Text Network Analysis is a sophisticated way to extract more meaning from the volumes and volumes of text based data we hold. This analysis emphasizes the frequency of mentioned words, but perhaps more importantly the relationships between these words, making for a much richer analysis.
This type of analysis is extremely useful when considering large volumes of verbatim text – such as employee feedback, customer emails and comments, news articles and other text based information sources. Text Network Analysis can be performed on any text. The words within the text are the “concepts” and we use their relationship to each other to define the links within the network. After the data is cleaned and pre-processed, we use specialist software (performed by the open-source Gephi software developed by Gephi Consortium) to produce the interactive network. The algorithm tests each concept against each other to determine the relationship between the concepts – which then influences the relative positioning of nodes to each other. The concepts that are connected (have a stronger relationship within the text) are pulled together, while the concepts that are not connected are pushed apart.
Using this type of analysis, we can identify and explore key themes, effectively sorting the concepts into the different groups according to how interconnected they are to one another. For example, in a recent analysis of customer feedback data, we coloured ‘satisfied’ concepts green and ‘unsatisfied’ concepts with red. Every other concept with a relationship to either of these key concepts became tinted with the same shade. Where a concept was related to both satisfied and unsatisfied, it became a mix of red and green (khaki).
The video below gives a high level overview of the power of this analysis.
Performance Analytics. Your data. Seen like never before.