ACL RD-TEC 1.0 Summarization of A94-1010
Paper Title:
IMPROVING LANGUAGE MODELS BY CLUSTERING TRAINING SENTENCES
IMPROVING LANGUAGE MODELS BY CLUSTERING TRAINING SENTENCES
Primarily assigned technology terms:
- algorithm
- approximation
- atis
- bigram language modeling
- clustering
- context modeling
- data collection
- database
- database access
- grouping
- hypothesis selection
- language modeling
- linguistic analysis
- maximum likelihood
- modeling
- n-best paradigm
- n-gram modeling
- parsing
- reasoning
- recognition
- recognizer
- speech recognition
- speech recognition and understanding
- speech recognizer
- speech understanding
Other assigned terms:
- acoustic information
- analogy
- approach
- backoff
- baseline score
- bigram
- bigram model
- case
- cluster
- clusters
- coefficient
- community
- concepts
- corpora
- correlations
- data sparseness
- dialogue model
- dialogues
- distribution
- entropy
- euclidean distance
- events
- experimental results
- fact
- grammar
- grammar rule
- grammar rules
- hypotheses
- hypothesis
- independence assumption
- interpretation
- knowledge
- language engine
- language model
- language models
- likelihood
- linguistic
- linguistic constraints
- linguistic information
- measure
- method
- model complexity
- model parameters
- n-gram
- n-gram model
- n-grams
- names
- natural language
- paragraph
- paragraphs
- parse
- parse tree
- parsing table
- perplexity
- precision
- probabilities
- probability
- probability distribution
- probability distributions
- process
- query
- relative frequency
- semantic
- semantic structures
- sentence
- sentences
- similarity measure
- size of the corpus
- subcorpus
- syntactic constructions
- syntax
- system performance
- technique
- test corpus
- test data
- training
- training corpus
- training data
- tree
- trigram
- trigram model
- understanding
- unigram
- unigram language model
- unigram probability
- user
- utterance
- vector space
- vocabulary
- word
- word classes
- word sequences
- word strings
- word-based evaluation
- words