news: 16 Newsgroups Dataset

In clintpgeorge/clda: Approximate Inference Algorithms for the Compound Latent Dirichlet Allocation Model

Description

It's a subset of the 20Newsgroups dataset. This corpus consists of 10,764 news articles and 9,208 unique words.

Usage

data("news")

Format

vocab a vector of unique words in the corpus vocabulary.

docs a list of documents in the corpus. Each item (represents a document) is a matrix (2 X U) of word frequencies, where U represents the number of unique words in a document. Each column in the matrix represents a unique word in a document and contains

• vocabulary-id. the index of the word in the vocabulary (starts with 0)

• frequency. the relative frequency of the word in the document

docs.metadata a matrix of document (article) metadata, where each row represents a document with

• category. the category assigned to the article

• name. the name of the news article from the 20Newsgroups dataset

• doclength. the number of words in the article

• collection. the collection name of each article

cids a vector of document collection IDs

class.labels a vector of categories (classes) in the corpus

collection.labels a vector of collections in the corpus

ds.name the corpus name (string)

num.docs the number of documents in the corpus

V the vocabulary size

Note

Created on July 26, 2015

Author(s)

Clint P. George

Source

Articles are downloaded via scikit-learn

See Also

Other datasets: nips, yelp

clintpgeorge/clda documentation built on May 13, 2019, 8 p.m.