潜在狄利克雷分配的R语言代码

以下是一个潜在狄利克雷分配的R语言代码，用于生成200个分布式的概率向量：

library(lda)
library(topicmodels)

set.seed(123)
K <- 5 # number of topics
V <- 50 # number of words
docs <- 200 # number of documents
alpha <- 0.1 # prior for document-topic distribution
eta <- 0.01 # prior for topic-word distribution

# generate random distribution of probability for each topic
theta <- matrix(rdirmulti(K, docs, alpha), docs)

# generate random distribution of probability for each word in each topic
phi <- matrix(rdirmulti(V, K, eta), V)

# simulate data based on topic and word probabilities
simdata <- LDA::simtopics(docs, phi, theta)

# fit LDA model
lda_model <- LDA(simdata$documents, K, method = "Gibbs", control = list(burnin = 500, iter = 500, thin = 1))

# extract estimated topic-word probabilities
beta_hat <- t(lda_model$topics) / lda_model$word.freq

# print estimated topic-word probabilities
print(beta_hat)

此代码生成200个文档的文本数据，并拟合一个具有5个主题的LDA模型，在概率向量中估计每个主题和每个单词。