ACL RD-TEC 1.0 Summarization of P05-1023
Paper Title:
DATA-DEFINED KERNELS FOR PARSE RERANKING DERIVED FROM PROBABILISTIC MODELS
DATA-DEFINED KERNELS FOR PARSE RERANKING DERIVED FROM PROBABILISTIC MODELS
Authors: James Henderson and Ivan Titov
Primarily assigned technology terms:
- algorithm
- beam search
- binary classification
- classification
- classifier
- classifier learning
- classifiers
- computing
- decoding
- dynamic programming
- dynamic programming approach
- error minimization
- estimator
- extractor
- fisher kernel
- history-based parsing
- inner product
- kernel
- kernels
- language parsing
- language processing
- learning
- learning algorithm
- left-corner parsing
- linear classifier optimization
- linear discriminant
- linear kernel
- loss function
- machine learning
- maximum entropy
- maximum likelihood
- natural language parsing
- natural language processing
- network training
- neural network
- neural network training
- nlp
- normalization
- optimization
- parameter estimation
- parameterization
- parse reranking
- parser
- parsers
- parsing
- parsing method
- perceptron
- perceptron algorithm
- perceptron training
- probabilistic parser
- processing
- reranking
- search
- statistical parser
- statistical parsers
- structured classification
- support vector machines
- training algorithm
- tree kernel
- validation
- voted perceptron
- voted perceptron algorithm
- voting
- word prediction
Other assigned terms:
- approach
- baseline model
- beam
- bias
- case
- classification error
- classification task
- classification tasks
- complete parse
- derivation
- distribution
- domain knowledge
- empirical results
- entropy
- entropy models
- estimation
- evaluation measure
- experimental results
- exponential model
- feature
- feature vector
- feature vectors
- finite set
- generative model
- generative probability
- generative probability model
- gold standard
- grammar
- heuristic
- history representation
- history-based model
- joint probability
- knowledge
- large training
- lexical tree
- likelihood
- linguistic
- linguistic knowledge
- log-likelihood
- mapping
- mappings
- maximum entropy models
- measure
- measures
- method
- model parameters
- natural language
- network architecture
- neural network architecture
- normalization factor
- parse
- parse tree
- parsing models
- parsing problem
- parsing strategy
- partial parse
- penn treebank
- phrase
- phrase structure
- phrase structure tree
- phrase structure trees
- posterior
- posterior probability
- precision
- probabilistic model
- probabilistic models
- probabilities
- probability
- probability distribution
- probability model
- process
- programming approach
- representations
- sentence
- sentence-tree pair
- sentences
- sigmoid function
- statistical models
- structural locality
- support vector
- synchrony networks
- technique
- term
- terms
- testing set
- trained model
- training
- training corpus
- training data
- training set
- tree
- tree adjoining grammar
- treebank
- treebank wsj corpus
- trees
- true probability distribution
- vocabulary
- word
- words
- wsj corpus
