| { | |
| "algorithm": { | |
| "command": null, | |
| "id": 0, | |
| "name": "Gensim Continuous Skipgram", | |
| "tool": "Gensim", | |
| "url": "https://github.com/RaRe-Technologies/gensim", | |
| "version": "3.6" | |
| }, | |
| "contents": [ | |
| { | |
| "filename": "model.txt", | |
| "format": "text" | |
| }, | |
| { | |
| "filename": "meta.json", | |
| "format": "json" | |
| } | |
| ], | |
| "corpus": [ | |
| { | |
| "NER": false, | |
| "case preserved": true, | |
| "description": "English Wikipedia Dump of February 2017", | |
| "id": 1, | |
| "language": "eng", | |
| "lemmatized": false, | |
| "public": true, | |
| "stop words removal": "NLTK", | |
| "tagger": null, | |
| "tagset": null, | |
| "tokens": 2252637050, | |
| "tool": "Wikipedia Extractor", | |
| "url": "https://dumps.wikimedia.org/" | |
| }, | |
| { | |
| "NER": false, | |
| "case preserved": true, | |
| "description": "Gigaword 5th Edition", | |
| "id": 4, | |
| "language": "eng", | |
| "lemmatized": false, | |
| "public": false, | |
| "stop words removal": "NLTK", | |
| "tagger": null, | |
| "tagset": null, | |
| "tokens": 4815382730, | |
| "tool": null, | |
| "url": "https://catalog.ldc.upenn.edu/LDC2011T07" | |
| } | |
| ], | |
| "creators": [ | |
| { | |
| "email": "andreku@ifi.uio.no", | |
| "name": "Andrey Kutuzov" | |
| } | |
| ], | |
| "dimensions": 300, | |
| "handle": "http://vectors.nlpl.eu/repository/20/18.zip", | |
| "id": 18, | |
| "iterations": 5, | |
| "vocabulary size": 291186, | |
| "window": 5 | |
| } |