自然語言處理（NLP）是人工智能領域的一個重要方向。NLP數據集具有行業相關性，如果你沒有可用的NLP數據集，希望如下NLP數據集對你有幫助。
1、Full-text corpus data
簡介：Full-text corpus data from six large corpora of English-- iWeb, NOW, Wikipedia, COCA, COHA, GloWbE
規模：
COCA（440 million words | 190,000 texts）
iWeb（13.8 billion words | 22 million web pages）
NOW（6.04 billion words | 6,000,000 texts）
Wikipedia（1.8 billion words | 4.4 million texts）
Corpus del Español (Web/Dialects)（1.8 billion words | 1,800,000 texts）
GloWbE（1.8 billion words | 1,800,000 texts）
COHA（385 million words | 115,000 texts）
地址：https://www.corpusdata.org/
2、Apache Software Foundation Public Mail Archives
簡介：A collection of all publicly available Apache Software Foundation mail archives as of July 11, 2011
規模：
地址：
https://aws.amazon.com/de/datasets/apache-software-foundation-public-mail-archives/
http://mail-archives.apache.org/mod_mbox/
3、WordNet
簡介：WordNet is a large lexical database of English. Nouns, verbs, adjectives and adverbs are grouped into sets of cognitive synonyms (synsets), each expressing a distinct concept
規模：117K synsets
地址：https://wordnet.princeton.edu/
4、The Blog Authorship Corpus
簡介：The Blog Authorship Corpus consists of the collected posts of 19,320 bloggers gathered from blogger.com in August 2004
規模：19,320 bloggers、681,288 posts、over 140 million words
地址：http://u.cs.biu.ac.il/~koppel/BlogCorpus.htm
5、Amazon product data
簡介：This dataset contains product reviews and metadata from Amazon
規模：142.8 million reviews
地址：http://jmcauley.ucsd.edu/data/amazon/
6、Web data: Amazon reviews
簡介：This dataset consists of reviews from amazon
規模：~35 million reviews
地址：https://snap.stanford.edu/data/web-Amazon.html
7、Common Crawl
簡介：A corpus of web crawl data composed of over 5 billion web pages
規模：5 billion web pages
地址：https://registry.opendata.aws/commoncrawl/
8、Yelp Open Dataset
簡介：The Yelp dataset is a subset of our businesses, reviews, and user data for use in personal, educational, and academic purposes
規模：5,996,996 reviews、188,593 businesses、280,992 pictures、10 metropolitan areas
地址：https://www.yelp.com/datasets
9、MACHINE TRANSLATION
簡介：Translation between language pairs
規模：30M
地址：http://statmt.org/wmt18/index.html
10、Enron Email Data
簡介：Enron email data publicly released as part of FERC's Western Energy Markets investigation converted to industry standard formats by EDRM
規模：1,227,255 emails with 493,384 attachments covering 151 custodians
地址：http://aws.amazon.com/de/datasets/enron-email-data/
11、Federal Contracts
簡介：A data dump of all federal contracts from the Federal Procurement Data Center found at USASpending.gov
規模：
地址：https://aws.amazon.com/de/datasets/federal-contracts-from-the-federal-procurement-data-center-usaspending-gov/
12、Sentiment140
簡介：Sentiment140 allows you to discover the sentiment of a brand, product, or topic on Twitter
規模：160K
地址：http://help.sentiment140.com/home
13、Freebase Data Dump
簡介：A data dump of all the current facts and assertions in the Freebase system
規模：millions of topics in hundreds of categories
地址：https://aws.amazon.com/de/datasets/freebase-data-dump/
14、IMDB movie review dataset
簡介：This is a dataset for binary sentiment classification containing substantially more data than previous benchmark datasets
規模：25K for training，25K for testing，additional unlabeled data
地址：http://ai.stanford.edu/~amaas/data/sentiment/
論文：http://www.aclweb.org/anthology/P11-1015
15、Web 1T 5-gram Version 1
簡介：Web 1T 5-gram Version 1, contributed by Google Inc., contains English word n-grams and their observed frequency counts
規模：1 trillion word tokens of text from publicly accessible Web pages
地址：https://catalog.ldc.upenn.edu/LDC2006T13
16、Harvard Library APIs & Datasets
簡介：World's largest academic libraries
規模：over 12.7M bib records、4M image records、2M finding aid components
地址：https://library.harvard.edu/services-tools/harvard-library-apis-datasets#Harvard-Library-Bibliographic-Dataset
17、Reddit comments may 2015
簡介：A small portion of the Reddit comments
規模：
地址：https://www.kaggle.com/reddit/reddit-comments-may-2015/home
18、Twenty Newsgroups Data Set
簡介：This data set consists of 20000 messages taken from 20 newsgroups
規模：20000 messages taken from 20 newsgroups
地址：https://archive.ics.uci.edu/ml/datasets/Twenty+Newsgroups
19、Crosswikis data
簡介：crosswikis-data
規模：
地址：https://nlp.stanford.edu/data/crosswikis-data.tar.bz2/
20、其他數據集
https://aws.amazon.com/de/datasets
https://www.kaggle.com/datasets
https://nlp.stanford.edu/links/statnlp.html
https://www.figure-eight.com/data-for-everyone
https://github.com/awesomedata/awesome-public-datasets
閱讀更多 深度學習社區 的文章
