metacurate_news_2022_1.json

{
  "params": {
    "visualize_top_n": 50,
    "report_top_n": 200,
    "cluster_probability": 0.75,
    "title": "Top AI/ML/data science and related news of 2022",
    "publish_to_plotly": true,
    "plotly_file_name": "metacurate_top_ai_ml_news_2022_1"
  },
  "data": {
    "raw": "./data/metacurate_news_2022.csv",
    "normalized": "./data/transient/normalized.csv",
    "clustered": "./data/transient/clustered.csv",
    "cluster_info": "./data/transient/cluster_info.csv",
    "cluster_descriptions": "./data/transient/cluster_descriptions.csv",
    "cluster_viz_data": "./data/output/2022_1/cluster_viz_data.csv",
    "cluster_viz_html": "./data/output/2022_1/metacurate_news_viz_2022.html",
    "cluster_report": "./data/output/2022_1/README.md",
    "cache": "./data/transient/.cache"
  },
  "resources": {
    "omit_strings": "./resources/omit_strings.csv"
  },
  "vectorizer": {
    "model_name_or_path": "all-mpnet-base-v2"
  },
  "clusterer": {
    "metric": "precomputed",
    "cluster_selection_method": "leaf",
    "min_cluster_size": 10,
    "min_samples": 2,
    "cluster_selection_epsilon":0.05,
    "memory": "./data/transient/.cache"
  }
}