I’ve been experimenting with 500,000 corporate emails and natural language processing for social network analysis.
Using graph algorithms we can infer information behaviour identifying clusters of ‘important’ people in a network based on incoming emails. This type of ‘importance’ might be hidden and not relate to formal roles and organisational hierarchy.
These can be further split by applying sentiment analysis to the email content, categorising email topics and classifying people external to the company. This could be used to determine ‘experts’ within the social network for certain topics. It may also highlight unconnected sub-networks where groups of people are discussing similar topics but are not in communication with one another.
#naturallanguageprocessing #knowledgegraph #knowledgemanagement
Leave a Reply