In network science, interpersonal relationships have traditionally been modeled with one type of edge. This is often an inadequate approximation of reality and does not capture the full nature of relationships. To address these shortcomings, researchers have identified ten social dimensions that can be used to describe the social character of relationships, and developed “Natural Language Processing”-tools The purpose of this thesis was to explore how the ten social dimensions can be applied in network science to capture the social character of interpersonal relationships when modeling social systems. The goal was to create a network that captures the social character of interpersonal relationships. We constructed a proof-of-concept by modeling the relationships between characters in the movie Les Miserables (1998).

Firstly, we constructed a dataset that contained the subtitle, speaker, receivers, and time of every utterance in the movie Les Miserables (1998). We then used Natural Language Processing to classify each utterance according to the ten social dimensions, and modeled characters and their verbal interactions with a Multilayer Temporal Text Network. From which we derived a Directed Weighted Multilayer Network representing the characters and the strength of their connections in each of the ten social dimensions. Finally, we generated weighted distributions that represented the Directed Weighted Multilayer Network’s representation of relationships between pairs of characters. These distributions were compared with a human’s perception of the relationships. The results showed that the distributions and the human’s perception either partially or largely matched for multiple relationships, indicating that the network captured key social characteristics of multiple relationships.