Background

#loading libraries
library(tidytext)
library(stringr)
library(tidyverse)
library(wordcloud)
library(dplyr)
library(reshape2)
library(lubridate)
library(ggplot2)

#load the rda file
load(file = "textdata.Rda")

#code to get started
#"pasted this part from Emily :), original code by Jeny :)"
df_tidy <- df_full %>% 
  unnest_tokens(word, content) %>% 
  anti_join(stop_words) 

Introduction

#word cloud of 100 most frequently written words
df_word <- df_tidy %>%
  count(word) %>%
  with(wordcloud(word, n, max.words = 100, color = palette()))

Inspired by how the Smith College administration uses email as the main communication to its wider community, we aimed to conduct text analysis of the number of emails the school has sent throughout the four years to assess its response to the COVID-19 pandemic. We believe that in a sea of emails filled with a lot of text, sentiment analysis makes sense because we need methods that can help us easily sort and understand the contents. In our analysis, we hope to contextualize the emails in three ways. In measuring the polarity of the emails, we will use a binary lexicon, bing, to classify the text as either negative or positive. We will also detect emotions within the text with the use of the nrc. Emotions are positive, negative, anger, anticipation, disgust, fear, joy, sadness, surprise, and trust. Lastly, we will assess the text numerically using a non binary, scaling method \([-5, 5]\) with the AFINN lexicon.