A Hybrid Approach to Adaptive Statistical Language Modeling

We describe our latest attempt at adaptive language modeling. At the heart of our approach is a Maximum Entropy (ME) model, which incorporates many knowledge sources in a consistent manner. The other components are a selective unigram cache, a conditional bigram cache, and a conventional static trigram. We describe the knowledge sources used to build such a model with ARPA's official WSJ corpus, and report on perplexity and word error rate results obtained with it. Then, three different adaptation paradigms are discussed, and an additional experiment, based on AP wire data, is used to compare them.

• Publication year

  1994

• Venue

  Human Language Technology - The Baltic Perspectiv

• Publication date

  1994-03-08

• Fields of study

  Computer Science

• Identifiers
  DOI 10.3115/1075812.1075827
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• 1993 Benchmark Tests for the ARPA Spoken Language Program

  1994cited by this paper
• Improving speech recognition performance via phone-dependent VQ codebooks and adaptive language models in SPHINX-II

  1994cited by this paper
• Adaptive Statistical Language Modeling; A Maximum Entropy Approach

  1994cited by this paper
• The Hub and Spoke Paradigm for CSR Evaluation

  1994cited by this paper
• An Overview of the SPHINX-II Speech Recognition System

  1993cited by this paper
• Adaptive Language Modeling Using the Maximum Entropy Principle

  1993cited by this paper
• The SPHINX-II speech recognition system: an overview

  1993cited by this paper
• Trigger-based language models: a maximum entropy approach

  1993cited by this paper
• Improvements in Stochastic Language Modeling

  1992cited by this paper
• A tree-based statistical language model for natural language speech recognition

  1989cited by this paper
• Speech Recognition and the Frequency of Recently Used Words: A Modified Markov Model for Natural Language

  1988cited by this paper
• Generalized Iterative Scaling for Log-Linear Models

  1972cited by this paper
• Information Theory and Statistical Mechanics

  1957cited by this paper

• Learning natural ordering of tags in domain-specific Q&A sites

  2021cites this paper
• Multi-timescale representation learning in LSTM Language Models

  2020cites this paper
• Automatic speech recognition: a survey

  2020cites this paper
• Developing an Adaptive Language Model for Bahasa Indonesia

  2019cites this paper
• Detecting Document-level Context Triggers to Resolve Translation Ambiguity

  2015cites this paper
• Learning Constructions of Natural Language: Statistical Models and Evaluations

  2012cites this paper
• Word prediction techniques for user adaptation and sparse data mitigation

  2011influential citation
• Word Prediction Techniques for User Adaptation and Sparse Data

  2008cites this paper
• Utilisation de la linguistique en reconnaissance de la parole : un état de l'art

  2006cites this paper
• Abbreviation Recognition with MaxEnt Model

  2006cites this paper
• Period Disambiguation with Maxent Model

  2005cites this paper
• Modeling long distance dependence in language: topic mixtures versus dynamic cache models

  1999cites this paper
• MAXIMUM LIKELIHOOD ESTIMATION OF MULTIPLE PRONUNCIATIONS FOR PROPER NOUNS

  1999cites this paper
• Dynamic Nonlocal Language Modeling via Hierarchical Topic-Based Adaptation

  1999influential citation
• EFFICIENT SEARCH ALGORITHMS FOR LARGE VOCABULARY CONTINUOUS SPEECH RECOGNITION

  1998cites this paper
• Category-Based Statistical Language Models

  1997cites this paper
• Methods of Category Classiication Applied to Word-sense Disambiguation and Discourse Analysis a Proposal Technical Information

  1996cites this paper
• A maximum entropy approach to adaptive statistical language modelling

  1996influential citation
• Incorporating diverse information sources in handwriting recognition postprocessing

  1996cites this paper
• Modeling long distance dependence in language: topic mixtures vs. dynamic cache models

  1996cites this paper
• Building Probabilistic Models for Natural Language

  1996influential citation
• A review of large-vocabulary continuous-speech

  1996cites this paper
• Automated speech understanding: the next generation

  1995cites this paper
• Algorithms for bigram and trigram word clustering

  1995cites this paper
• Quantifying lexical influence: Giving direction to context

  1995cites this paper
• Methodologies for language modeling and search in continuous speech recognition

  1995cites this paper
• Training and Application of Integrated Grammar/Bigram Language Models

  1994cites this paper
• Word-Sense Disambiguation Using Decomposable Models

  1994cites this paper
• Word-Sense Disambiguation Using Decomposable Models

  1994cites this paper
• Analyzing and Improving Statistical Language Models for Speech Recognition

  1994cites this paper
• Adaptive Statistical Language Modeling; A Maximum Entropy Approach

  1994cites this paper
• Improving Language Models by Clustering Training Sentences

  1994cites this paper
• 1993 Benchmark Tests for the ARPA Spoken Language Program

  1994cites this paper
• Boston University College of Engineering Thesis Language Modeling with Sentence-level Mixtures

  year unknowncites this paper