This page introduces the story and methodology behind our mining memory visualizations. These tools uncover 90 years of Nigerian media reports to restore forgotten voices, map silences, and provoke new inquiry.
These visualizations were not just built, they were felt. Each one carries a thread of memory, from the children lost in the 1943 Enugu coal collapse to contemporary debates on illegal gold mining in Zamfara. Inspired by Dr. Kalani Craig’s advice on genre, tokenization, and NER workflows, our team used natural language processing to lift recurring names, places, and events from the archive and represent them interactively.
Field notes, archival scans, and spreadsheet tables became maps and models. The team iterated through errors, misspellings, fuzzy dates, overlapping locations, to ensure our storytelling remained ethical, evocative, and true to lived histories.
Description: A map-based visualization geolocating and clustering 90 years of mining-related newspaper reports. Each cluster reveals a moment of tension, dispossession, resistance, or state-corporate complicity.
Method: Folium map + GeoJSON with Leaflet clustering
Tools Used: Python, Pandas, Folium
How to Use: Click on a pin to open a pop-up with title, source, and date.
Audience: Historians, environmental researchers, policy analysts
Unique Insight: Reveals where extractive memory is concentrated geographically.
Description: Shows the yearly distribution of mining articles with peaks linked to disasters, protests, and regulatory change (e.g., 1943 child deaths, 2006 collapse, 2020 reforms).
Method: Bokeh + hover tools and sliders
Tools Used: Bokeh, JavaScript, Pandas
How to Use: Hover for article snippets. Filter by time.
Audience: Media scholars, legal historians, educators
Unique Insight: Connects frequency to tragedy and reform moments.
Description: Topic modeling from 1935–2025 that visualizes themes like forced labor, pollution, nationalism, and memory politics. It shows how the narrative moved from community-driven to tragedy-centered over time.
Method: LDA topic modeling over tokenized text
Tools Used: Gensim, spaCy, Plotly
How to Use: Select a decade to explore dominant themes.
Audience: Memory scholars, communication researchers
Unique Insight: Traces thematic shifts tied to governance and resource control.
Description: Focuses on densely clustered areas like Enugu and Jos, enabling granular exploration of community narratives, especially where multiple reports originate from one site.
Method: Leaflet time sequencing + spatial separation
Tools Used: JavaScript, Leaflet, GeoJSON
How to Use: Click individual markers to explore event-level data.
Audience: Local historians, spatial researchers
Unique Insight: Adds spatial clarity and temporal sequence in tight geographies.
This dashboard reveals which people, places, and organizations were named, and which were erased. It’s a tool for surfacing power, silence, and attention.
How to Use: Search “date”, “Organizations”, or “persons”. Compare their visibility over time.
Built With: spaCy NER models, Streamlit frontend.
Audience: Data scientists, human rights advocates, memory researchers.
The visualizations link directly to our Echefùla Blog and Commons Forum, where users can share reflections, write ethnographies based on articles, or request archival copies. Together, we aim to build a participatory data commons where visual evidence meets personal memory.
Built by Uzoamaka Nwachukwu · Visualization design supported by IDAH and SeekCommons · 2025