Classify — Text & Data Classification Tool

A text and data classification skill supporting keyword-based and TF-IDF classification methods. Train models on labeled data, predict categories for new text, evaluate accuracy, and manage labeled datasets. All data stored in JSONL format.

Prerequisites

• bash (v4+)
• python3 (v3.6+)
• No external dependencies required (uses built-in collections and math modules)

Environment Variables

Data Storage

• Training data: ~/.classify/data.jsonl
• Models: ~/.classify/models/
• Config: ~/.classify/config.json
• Predictions log: ~/.classify/predictions.jsonl

Commands

train

Add labeled training data to a model.

CLASSIFY_TEXT="Great product, love it!" CLASSIFY_LABEL="positive" CLASSIFY_MODEL="sentiment" scripts/script.sh train

predict

Classify new text using a trained model.

CLASSIFY_TEXT="This is terrible" CLASSIFY_MODEL="sentiment" scripts/script.sh predict

evaluate

Evaluate model accuracy with test data.

CLASSIFY_MODEL="sentiment" scripts/script.sh evaluate

label

Add or update a label for existing data.

CLASSIFY_ID="cls_abc123" CLASSIFY_LABEL="neutral" scripts/script.sh label

list

List training data or prediction history.

CLASSIFY_MODEL="sentiment" CLASSIFY_LABEL="positive" scripts/script.sh list

export

Export model data or predictions.

CLASSIFY_MODEL="sentiment" CLASSIFY_FORMAT="csv" scripts/script.sh export

import

Import labeled data from a file.

CLASSIFY_FILE="/path/to/labeled_data.csv" CLASSIFY_MODEL="sentiment" scripts/script.sh import

stats

Show model statistics and label distribution.

CLASSIFY_MODEL="sentiment" scripts/script.sh stats

config

View or update classification configuration.

CLASSIFY_KEY="default_method" CLASSIFY_VALUE="tfidf" scripts/script.sh config

batch

Classify multiple texts from a file.

CLASSIFY_FILE="/path/to/texts.txt" CLASSIFY_MODEL="sentiment" scripts/script.sh batch

help

Display usage information.

scripts/script.sh help

version

Display current version.

scripts/script.sh version

Output Format

{
  "status": "success",
  "command": "predict",
  "data": {
    "text": "This product is amazing",
    "predicted_label": "positive",
    "confidence": 0.85,
    "model": "sentiment"
  }
}

Error Handling

Powered by BytesAgain | bytesagain.com | hello@bytesagain.com