ACL RD-TEC 1.0 Summarization of P99-1022
Paper Title:
DYNAMIC NONLOCAL LANGUAGE MODELING VIA HIERARCHICAL TOPIC-BASED ADAPTATION
DYNAMIC NONLOCAL LANGUAGE MODELING VIA HIERARCHICAL TOPIC-BASED ADAPTATION
Authors: Radu Florian and David Yarowsk
Primarily assigned technology terms:
- adaptive topic-probability estimation
- agglomerative clustering
- algorithm
- bottom-up clustering
- classification
- classifiers
- clustering
- clustering algorithm
- collapsing
- comparative analysis
- computing
- constrained optimization
- distance function
- document clustering
- em-based clustering
- error rate reduction
- hierarchical clustering
- hierarchical interpolation
- hierarchical smoothing
- interpolation algorithm
- k-means
- k-means clustering
- k-nn
- language modeling
- maximum entropy
- model construction
- model estimation
- model interpolation
- modeling
- naive bayes
- naive bayes classifiers
- normalization
- optimization
- parameter re-estimation
- probability estimation
- probability reestimation and interpolation
- rate reduction
- re-estimation
- reestimation
- smoothing
- topic adaptation
- topic detection
- topic-detection
- topic-probability estimation
- tree construction
- tree generation
- unsupervised algorithm
Other assigned terms:
- approach
- baseline model
- bigram
- bigram model
- broadcast news
- broadcast news corpus
- cache
- case
- cluster
- clustering procedure
- clusters
- community
- content words
- cosine similarity
- discourse
- distribution
- document
- document similarity
- entropy
- error rate
- estimation
- evaluation function
- evaluations
- events
- function words
- generation
- hierarchical structure
- histogram
- hypotheses
- hypothesis
- implementation
- intention
- inter-cluster distance
- intercluster similarity
- interpolation
- language model
- language models
- leaf
- measures
- mechanisms
- method
- model combination
- model probability
- news corpus
- noise
- optimization problem
- perplexity
- perplexity reduction
- probabilities
- probability
- probability estimate
- procedure
- process
- relative frequency
- run-time
- similarity measures
- size of the corpus
- speech corpus
- subtree
- switchboard corpus
- target vocabulary
- technique
- term
- terms
- text
- topic variation
- topic-probability
- topics
- training
- training data
- training phase
- transcripts
- tree
- tree structure
- trees
- trigram
- trigram model
- unigram
- unigram model
- vocabulary
- vocabulary size
- word
- word error rate
- word sequence
- word usage
- words
