Paste text or load an existing questionnaire from the database
Questionnaire Items
Original
English
Saved Questionnaires
Loadingβ¦
Build JSON
Configure metadata and generate the questionnaire structure
Questionnaire Metadata
Click a preset to load it, or type your own
JSON Previeweditable
β Insert into MongoDB
Loaded Questionnaire
Label Table
Click any label to add to Final Β· use + to pick from collection
Original Text
English Text
Original Labels
Semantic Labels
Semantic LLM
Topic Labels
Final Labels
π
Labels will load automatically
Update Questionnaire
Final labels are merged as topics into each question β review and save
JSON Previeweditable
Topic Selection
Build your topic set from the taxonomy, then filter by importance, urgency and severity
Selected Topics
Taxonomy Tree
Important Topics
Urgent (LLM)
Urgent (from DB, urgent=true)
Severity per Topic (0 β 1)
LLM Proposals (click Apply to use)
New Custom Labels to Insert
These labels were added during labeling but are not yet in the database. Insert them now.
Final Topics JSONeditable
BotBank Analysis
Generate direct questions from questionnaire items and match against BotBank
Filter by Project
π€
Click Generate to populate
π€
Click Generate to populate
Upload Question File
CSV format: two columns β text (question string) and topics (list of strings, e.g. ['sleep', 'mood'] or plain sleep,mood).
First row must be a header: text,topics. Duplicate questions are removed automatically (first occurrence kept).
Generation Prompts
π
Upload a .txt file to begin
π
Loadingβ¦
Saved Questionnaires
Browse questionnaires stored in the app database
β³
Loadingβ¦
Questions
Subject
Question
Default Options
Raw JSON
Add Labels
New labels go to the pending review queue and can be promoted to the main collection later.
UserBank
Statistics from the UserBank collection
π Persona
π₯
Loadingβ¦
π Persona
Load questions for a project
π¬
Select a project to load its accepted questions
Upload Labels CSV
Required columns: db_id, text, plus any number of label columns
whose values are lists β e.g. ['fatigue', 'sleep'].
Label columns are auto-detected. All columns other than db_id and text are treated as label sources.
Load from Database
Select a label β fetches all UserBank items containing it β Original Labels column auto-populated.
Label Columns β select which contribute to Golden
Label Statistics
β‘ labels & β matrix toggles are per-row
Precision β of predicted labels, how many are correctRecall β of golden labels, how many were predictedF1 β harmonic mean of precision & recallExact match β rows where predicted set = golden set exactly