EvergreenLP: Culturally Relevant Data Science via the #BlackGirlMagic Lens

TL;DR

EvergreenLP is an innovative learning platform that uses social media—specifically Twitter/X—as a living laboratory for data science. By focusing on the #BlackGirlMagic movement, researchers at Spelman College created a framework where African-American female students learn data extraction, visualization, and deconstruction through a lens that is culturally meaningful to them.

Academic Positioning: This work bridges the gap between Critical Media Literacy and Computational Social Science, moving beyond simple "coding bootcamps" toward a holistic, constructivist approach to data literacy.

Problem & Motivation: The Gap in Data Science Education

Despite the explosion of Big Data, the techniques to collect, store, and analyze it are often omitted from the undergraduate experience unless a student is a Computer Science major. Furthermore, the "Digital Divide" isn't just about access to hardware; it's about the transition from being a technological consumer to a technological producer.

The authors argue that students are already immersed in social media, yet they lack the tools to critically analyze the data they generate. For African-American students, whose voices have historically been silenced, "Black Twitter" serves as a vital tool for social change. The motivation here is to use these "digital artifacts" (hashtags like #BlackLivesMatter or #BlackGirlMagic) as the primary instructional material for data science.

Methodology: The EvergreenLP Framework

The platform employs a rapid prototyping paradigm. It allows Computer Science students to build the "back-end" (API connections and database management) while non-computing students use the "front-end" to perform interdisciplinary analysis.

1. Critical Media Literacy (CML)

The pedagogy rests on five core components:

• Authorship: Who created the content?
• Format: How does the platform influence the message?
• Audience: Is the content specific to a sub-community?
• Content: What values are embedded?
• Purpose: Does this promote profit, power, or positive narratives?

Figure 1: The interface allows students to query hashtags and keywords to filter raw social media data.

2. Systematic Deconstruction and Analysis

Students follow a workflow of Extraction → Deconstruction → Analysis → Construction. They analyze word clouds, language frequencies, and location data to understand the global reach of their community's digital presence.

Figure 2: Built-in visualization tools help students derive insights into language and sentiment without needing deep coding knowledge initially.

Experiments & Results: #BlackGirlMagic in Action

The study involved 17 first-year students at Spelman College from various majors (Biology, English, Psychology, etc.).

Key Findings:

• Beyond the Surface: Students initially felt "comfortable" with data but lacked advanced skills. After using EvergreenLP, they were able to perform Co-occurrence Network Analysis.
• Thematic Discovery: By analyzing their own essays on #BlackGirlMagic, students identified three core classifications: Internalization, Expression, and Expansion.
• Unexpected Insights: One student noted an anomaly where 401 tweets in their fetch were in Japanese, leading to a cross-cultural inquiry into how Black culture is perceived in Asian contexts.

Figure 3: A co-occurrence network representing how students define 'Greatness' and 'Movement' within the #BGM context.

Critical Analysis & Conclusion

EvergreenLP proves that culturally relevant data is a powerful hook for STEM engagement. By treating hashtags as "valuable artifacts of Black culture," the authors successfully taught Regression, Clustering, and Classification (Table II in the paper) to a non-technical audience.

Takeaway & Future Work

The core value of this work lies in its inclusive architecture—making the complex "invisible" while keeping the analysis "visible" and relevant. However, the study is early-stage; future iterations will need to address the "API fatigue" caused by changing social media access policies and integrate more automated sentiment analysis tools.

Final Insight: When students find themselves inside the data, the motivation to learn the science behind it becomes organic rather than forced.