Documentation Index
Fetch the complete documentation index at: https://docs.praison.ai/llms.txt
Use this file to discover all available pages before exploring further.
AI News Deduper
Deduplicate news articles using semantic similarity and cluster them by topic.
CLI Quickstart
# Run deduplication
praisonai recipe run ai-news-deduper \
--input '{"articles": [...], "similarity_threshold": 0.85}' \
--json
Use in Your App (SDK)
from praisonai.recipe import run, run_stream
# Deduplicate articles
result = run(
"ai-news-deduper",
input={
"articles": articles_list,
"similarity_threshold": 0.85,
"use_semantic": True
}
)
# Direct tool usage
import sys
sys.path.insert(0, 'agent_recipes/templates/ai-news-deduper')
from tools import deduplicate_articles, cluster_by_topic
# Deduplicate
deduped = deduplicate_articles(articles, similarity_threshold=0.85)
# Cluster by topic
clusters = cluster_by_topic(deduped["deduplicated"], num_clusters=5)
{
"type": "object",
"properties": {
"articles": {
"type": "array",
"description": "List of article objects"
},
"similarity_threshold": {
"type": "number",
"default": 0.85
},
"use_semantic": {
"type": "boolean",
"default": true
}
}
}
Output Schema
{
"deduplicated": [{"title": "...", "url": "..."}],
"removed_count": 5,
"clusters": [
{"topic": "GPT-5", "articles": [...]}
]
}
Configuration
| Option | Type | Default | Description |
|---|
| similarity_threshold | float | 0.85 | Similarity threshold for deduplication |
| use_semantic | bool | true | Use semantic similarity (requires OpenAI) |
| num_clusters | int | 5 | Number of topic clusters |
Dependencies
Environment Variables
| Variable | Required | Description |
|---|
| OPENAI_API_KEY | Yes | For semantic embeddings |
