id	author	title	date	pages	extension	mime	words	sentences	flesch	summary	cache	txt
work_yiggwirfrrdmxotzboejxtmxuy	Dani Yogatama	Dynamic Language Models for Streaming Text	2014	12	.pdf	application/pdf	7111	905	74	Dynamic Language Models for Streaming Text Language models are typically assumed to be static—the word-given-context Our model also exploits observable context variables to capture temporal variation that is otherwise provide useful auxiliary information that might indicate the similarity of language models across different timesteps. Language modeling for streaming datasets in the context of machine together temporal dynamics, conditioning on nonlinguistic context, and scalable online learning suitable for streaming data and extensible to include to learn language models for streaming datasets. The intuition behind the model is that the probability of a word appearing at day t depends on the background log-frequencies, the deviation coefficients of the word at previous timesteps β1:t−1, and the similarity of current conditions of the world (based on online learning of dynamic topic models. one," and "base exp" are unigram language models. one," and "base exp" are bigram language models. dataset with bigram base models, the five stocks with	./cache/work_yiggwirfrrdmxotzboejxtmxuy.pdf	./txt/work_yiggwirfrrdmxotzboejxtmxuy.txt