-
Notifications
You must be signed in to change notification settings - Fork 0
Smishing dataset
Bell Eapen edited this page Nov 10, 2025
·
2 revisions
Walkthrough for an SMS phishing dataset focusing on text-first analysis with optional numeric features.
- Prepare and import
mkdir smishing_demo && cd smishing_demo
mkdir crisp_source
# Copy dataset CSV to crisp_source
crisp --source crisp_source --out crisp_input --unstructured text- Qualitative analysis
crisp --inp crisp_input --assign --out crisp_input
crisp --inp crisp_input --print "documents metadata"- Semantic filtering examples
crisp --inp crisp_input --filters keywords=bank
crisp --inp crisp_input --filters keywords=otp- Visualization
crispviz --inp crisp_input --wordcloud --out viz_out/- Optional quantitative steps (if labels available)
crisp --inp crisp_input --ml --cart --regression --out crisp_input- Examine themes such as urgency and impersonation; compare with quantitative predictors (e.g., message length, presence of URLs) if available (Mettler et al., 2025).
- Use TDABM to identify structural neighborhoods for targeted memo writing (Rudkin & Dlotko, 2024).
This page documents a smishing (SMS phishing) dataset walkthrough analogous to the COVID demos, focusing on text-first analysis and optional numeric integration where labels exist.
- Prepare folders and import:
mkdir smishing_demo && cd smishing_demo
mkdir crisp_source
# Copy dataset CSV to crisp_source
crisp --source crisp_source --out crisp_input --unstructured text- Qualitative analysis:
crisp --inp crisp_input --assign --out crisp_input
crisp --inp crisp_input --print "documents metadata"- Semantic filtering examples:
crisp --inp crisp_input --filters keywords=bank
crisp --inp crisp_input --filters keywords=otp- Visualization:
crispviz --inp crisp_input --wordcloud --out viz_out/- Optional quantitative steps (if labels available):
crisp --inp crisp_input --ml --cart --regression --out crisp_input- Examine themes (e.g., urgency, impersonation) and relate to numeric predictors (e.g., message length) when available.
- Use semantic chunk search for exemplar phrases to refine categories (e.g., "verify account").