Skip to content

Interaction-Partitioned Topic Models (IPTM) using a Point Process Approach

Notifications You must be signed in to change notification settings

desmarais-lab/IPTM

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Functions jointly analyze the history of interaction (between the sender and receiver) and the contents or topic of text (e.g. email, online messages) using Markov chain Monte Carlo (MCMC).

The subdirectory pkg contains the actual package. The package can be installed with devtools.

devtools::install_github("desmarais-lab/IPTM", subdir = "pkg")

Functionality includes:

  • MCMC which runs Markov chain Monte Carlo (MCMC) algorithm to compute the time-weighted network statistics from the point process model of Perry and Wolfe (2013) and token-topic assignments of the documents over the corpus, given the interaction-pattern (IP) assignment of each document.

  • TableBetaIP and plot_betaIP which summarize the MCMC chain of network statistics for each interaction-pattern (IP), by drawing table and boxplot, respectively.

  • PlotTopic and PlotTopicIP which plot the topic distributions without (former) and with (latter) considering the interaction-pattern (IP) assignments.

  • TableWordIP which generates the table that summarizes interaction-pattern (IP) and token-topic assignments. For each interaction-pattern (IP), the topics of highest proportion and their corresponding most likely words are included.

The last three functions above are used to make comparison between different interaction-patterns (IP).

Releases

No releases published

Packages

No packages published