Week 4: Lexicon-Based Sentiment Analysis

Lexicon-based sentiment analysis applied to Bluesky social media data.

Learning Objectives

By the end of this week, you will be able to:

Authenticate with a social media API and retrieve posts by keyword using {atrrr}
Clean and tokenize social media text for downstream sentiment analysis
Apply and interpret multiple lexicon-based sentiment methods
Identify common failure modes of general-purpose lexicons on domain-specific environmental text
Compare method outputs using correlation matrices and distribution visualizations
Evaluate which method is most appropriate for a given corpus and research question

Lecture Materials

Slides

🖥 Slide deck from lecture

View Slides

Code & Data

💻 R scripts used during class:

Lab 3 Markdown

📦 Packages

New packages introduced this week:

Package	Purpose	Install
`atrrr`	Bluesky (AT Protocol) API client — search posts, authenticate, retrieve feeds	`install.packages("atrrr")`
`sentimentr`	Sentence-level sentiment scoring with valence shifters (negation, amplifiers)	`install.packages("sentimentr")`
`tidyvader`	Tidy interface to the VADER sentiment lexicon, optimized for social media text	`remotes::install_github("chris31415926535/tidyvader")`

📂 Assignments

Lab 3 — Lexicon-Based Sentiment Analysis with Bluesky Data Graded · 11 pts

Due: April 28 @ 11:59 pm

Complete the assignment laid out in the Lab 3 markdown. Submit your .Rmd and knitted HTML to Canvas. Make sure your Bluesky credentials are stored in .Renviron and do not appear anywhere in your submitted file. Look your knitted product over before submission to make sure all plots rendered and extraneous print-outs did not render.