/visualising-film-networks

A shiny implementation of the film networks analysis

Primary LanguageR

Visualising Film Networks

A shiny implementation of the film networks analysis. This is a web app that I built to complement my analysis of how films have changed across the decades from the perspective of male and female characters. The networks is a word co-occurrence networks built from movie plots scraped from wikipedia using the 'rvest' library. The network for each decade is built from 1000 movie plots, consisting of 100 movie plots from each year. Plots were parsed into tokens with parts of speech ('pos') tagged to them using spacyr, which runs with spacy python at the backend. 'GenderizeR' was then used to categorise the gender of the male and female entities that were extracted. After this, networks were constructed with the male and female characters as The central node by connecting co-occurring words in their order of the log likelihood of their co-occurrence. The green nodes are the immediate co-occurrences of all male/female characters in the plots. The orange nodes are words that co-occur with the green nodes, and are hence secondary co-occurrences.