[

概要:

In this paper, we investigate statistical network grammar using Ergodic Hidden Markov Model(HMM).

There are two types of natural language modeling. One is the class of deterministic models like network grammar or context free grammar, that exploit some known specific properties of the language, and the other is the class of statistical models like bigram or trigram in which one tries to characterize the statistical properties of the corpus.These statistical models include stochastic context free grammar and Markov process, a sort of non-deterministic finite state automaton. A HMM effectively exploits language model as a random process. By choosing specific parameters of HMM, grammatical rule can be estimated in a well-defined manner as a transition network. HMM is very rich in mathematical structure so that language models are determined more precisely than that of stochastic network grammar or Markov process.

This paper includes the results obtained to statistical network grammar automatically from about 4000 words using Ergodic HMM. The resultant model indicates that some grammatical features exist even though the process is carried out automatically.

Ergodic HMMによる確率つきネットワーク文法の獲得の可能性について
The Possibility for Acquisition of Statistical Network Grammar using Ergodic HMM

村上仁一	山本寛樹	嵯峨山茂樹
Jin'ichi Murakami	Hiroki Yamatomo	Shigeki Sagayama

ATR 自動翻訳電話研究所
ATR Interpretating Telephony Research Laboratories

]

論文をps形式でダウンロードする (約1Mbyte)