Czech Journals Pretrained Models (Memokre/spravodajske_weby)

This repository serves as the official model and asset store for the Czech Journals Orange3 Add-on.

Structure

.
├── RAW_VECTOR_MODELS/
│   ├── glove/
│   │   ├── glove.bin
│   │   ├── glove_vectors.txt
│   │   └── vocab.txt
│   └── word2vec/
│       ├── word2vec.bin
│       ├── word2vec_final.model
│       ├── word2vec_final.model.syn1neg.npy
│       ├── word2vec_final.model.wv.vectors.npy
│       └── word2vec_vectors.txt
└── TOPIC_MODELING/
    ├── CTM/
    │   └── CTM.bin
    ├── LDA/
    │   └── LDA.bin
    ├── PAM/
    │   └── PAM.bin
    ├── GloVe/
    │   ├── GloVe.bin
    │   └── topic_centroids.npy
    └── Word2Vec/
        ├── Word2Vec.bin
        └── topic_centroids.npy

Intended Usage

These pretrained weights and vector files are automatically fetched or loaded by Orange3 Czech Journals widgets:

• Topic Map Labeler (inferring topic distributions with LDA / PAM / CTM / Word2Vec / GloVe)
• Topic Distribution Diagram
• Topic Matrix
• Embedding vectorizers and corpus cleaners

All human-readable topic maps, dictionary mappings, and CSV/JSON reports reside in the companion GitHub repository under Addons/Czech_journals/DATA/topicmaps/.