Security Analysis Dashboard
National linguistic analysis of offensive and hateful social media content
Multilingual
Total Corpus Keywords
25504
Hate Speech Flags
1582
Abusive & Offensive
19700
Clean / Unoffensive
4222
Zambian Multilingual Cyber Security
Social media platforms in Zambia experience significant text volume across local languages. Standard English models fail to detect highly offensive slurs and tribalistic hate speech posted in local dialects like Chibemba, Chinyanja, Chitonga, or Silozi.
CyberDetect bridges this technological gap by providing custom text preprocessing and class balancing mechanisms (oversampling/undersampling) to train high-fidelity ML models (Logistic Regression + TF-IDF) specialized in Zambian dialects.
Supported Local Dialects:
🇿🇲
Chibemba (Bemba)
Northern / Copperbelt
🇿🇲
Chinyanja (Nyanja)
Lusaka / Eastern
🇿🇲
Chitonga (Tonga)
Southern Province
🇿🇲
Silozi (Lozi)
Western Province
Active Local Models
4 Active
Chinyanja Model
Acc
F1: Naive Bayes Classifier (Random Forest)
Trained: 2026-08-17 13:44
All Model
Acc
F1: Decision Tree
Trained: 2026-06-01 13:45
Chibemba Model
Acc
F1: Logistic Regression
Trained: 2026-06-01 12:35
English Model
Acc
F1: Logistic Regression
Trained: 2026-06-01 12:42