Text Analysis

Python Text Analysis: Word Embeddings

October 25, 2023, 2:00pm
How can we use neural networks to create meaningful representations of words? The bag-of-words is limited in its ability to characterize text, because it does not utilize word context.

Python Text Analysis: Topic Modeling

March 29, 2023, 2:00pm
In this part, we study unsupervised learning of text data. This is a stand alone work that builds from the two-part text analysis series.

Tactics for Text Mining non-Roman Scripts

April 15, 2024
Hilary Faxon, Ph.D.
by Hilary Faxon, Ph.D. & Win Moe. Non-Roman scripts pose particular challenges for text mining. Here, we reflect on a project that used text mining alongside qualitative coding to understand the politicization of online content following Myanmar’s 2021 military coup.

Addison Pickrell

IUSE Undergraduate Advisory Board
Mathematics
Sociology

Addison is an aspiring mathematician and social scientist (Class of '27). He loves collecting books he'll never read, is an open-source and open-access advocate, and an aspiring community organizer and systems disrupter. Ask me about community-based participatory action research (CBPAR), critical pedagogy, applied mathematics, and social science.

María Martín López

Data Science Fellow
Psychology

María Martín López is a PhD student in the Cognition area within the Department of Psychology. Her research relates to cognitive computational and quantitative models of individual differences in behaviors, thoughts, and emotions. She is particularly interested in how we can create and leverage novel algorithms to understand, measure, and predict processes relating to externalizing psychopathology (e.g. impulsivity, aggression, substance use). She answers these questions using a range of computational and quantitive models including AI, NLP, SEM, time series analysis, multi-level...

Twitter Text Analysis: A Friendly Introduction, Part 2

March 7, 2023
Mingyu Yuan
by Mingyu Yuan. This blog post is the second part of “Twitter Text Analysis”. The goal is to use language models such as BERT to build a classifier on tweets. Word embedding, training and test splitting, model implementation, and model evaluation are introduced in this model.

Why We Need Digital Hermeneutics

July 13, 2023
Tom van Nuenen
by Tom van Nuenen. Tom van Nuenen discusses the sixth iteration of his course named Digital Hermeneutics at Berkeley. The class teaches the practices of data science and text analysis in the context of hermeneutics, the study of interpretation. In the course, students analyze texts from Reddit communities, focusing on how these communities make sense of the world. This task combines both close and distant readings of texts, as students employ computational tools to find broader patterns and themes. The article reflects on the rise of AI language models like ChatGPT, and how these machines interpret human interpretations. The popularity and profitability of language models presents an issue for the future of open research, due to the monetization of social media data.

Unlock the Joy and Power of Reading in Language Learning

August 21, 2023
Bowen Wang-Kildegaard
by Bowen Wang-Kildegaard. I share my story of how reading for pleasure transformed my English speaking and writing skills. This experience inspired my passion to promote the joy and power of reading to all language learners. Using natural language processing techniques, I dive into the Language Learning subreddit, revealing a trend: Learners are often highly anxious about output practices, but are generally positive about input methods like reading and listening. I then distill complex language learning theories into actionable language learning tips, emphasizing the value of extensive reading for pleasure, pointing to potential methods like using ChatGPT for customization of reading materials, and advocating for joy in the learning journey.

My Summer Exploring Data Science for Social Justice: Learnings, Tensions & Recommendations

September 5, 2023
Genevieve Smith
by Genevieve Smith. This summer I joined the D-Lab hosted Data Science for Social Justice workshop at UC Berkeley diving into Python – including TF-IDF, sentiment analysis, word embeddings, and more – with a lens towards leveraging data science for social justice. My team explored a Reddit channel on abortion and used computational analysis to answer key questions related to abortion access from before versus after Roe vs. Wade was overturned. Computational social science is incredibly powerful, but I continue to grapple with tensions particularly as it relates to employing machine learning and large language in international research, and end with key recommendations for CSS practitioners.

Twitter Text Analysis: A Friendly Introduction

October 25, 2022
Mingyu Yuan

Read part 2 here.

Introduction

Text analysis techniques, including sentiment analysis, topic modeling, and named entity recognition, have been increasingly used to probe patterns in a variety of text-based documents, such as books, social media posts, and others. This blog post introduces Twitter text analysis, but is not intended to cover all of the aforementioned topics. The tutorial is broken down into two parts. In this very first post, I...