Natural Language Understanding

1
Natural Language Understanding

• Raivydas Simenas

2
Overwiev

• History
• Speech Recognition
• Natural Language Understanding
• statistical methods to resolve ambiguities
• Current situation

3
History

• Roots in teaching the deaf to speak using
  visible speech
• 1874 Alexander Bells invention of harmonic
  telegraph
• Different frequency harmonics from an electrical
  signal could be separated
• Could sent multiple messages over the same wire
  at the same time
• 1940s separating the speech signal into
  different frequency components using the
  spectrogram
• 1950s the beginning of computer use for
  automatic speech recognition

4
The Nature of Speech

• Phoneme a basic sound, e.g. a vowel
• The complexity of human vocal apparatus about 18
  phonemes per second
• Speech viewed as a sound wave
• Identifying sounds analyzing the sound wave into
  its frequency components

5
The Spectrogram I

• A visual representation of speech which contains
  all the salient information
• Plots the amount of energy at different
  frequencies against time
• Discontinuous speech (making a pause after each
  word) easier to recognize on the spectrogram

6
The Spectrogram II

• The same word uttered twice (especially by
  different speakers speaker independence) might
  look radically different on a spectrogram
• The need to recognize invariant features in a
  spectrogram
• Formants resonant frequencies sustained for a
  short time period in pronouncing a vowel
• Normalization distinguishing between relevant
  and irrelevant information
• Nonlinear time compression taking care of the
  changing speed of a speech
• Matching a spoken word to a template

7
Robust Speech Recognition

• Need to maintain accuracy when the quality of the
  input speech is degraded or when the speech
  characteristics differ due to change in
  environment or speakers
• Dynamic parameter adaptation either alter the
  input signal or the internally stored
  representations
• Optimal parameter estimation based on a
  statistical model characterizing the differences
  between training and test sets
• Empirical feature comparison based on comparison
  between high-quality speech and the same speech
  recorded under degraded conditions

8
Stochastic Methods in Speech Recognition

• Generating the sequence of word hypotheses for an
  acoustic signal is most often done using
  statistics
• The process
• A sequence of acoustic signals is represented
  using a collection of vectors
• Such collections are used to build acoustic word
  models, which consist of probabilities of certain
  sequences of vectors representing a word
• Acoustic word models utilize Markov chains

9
Representing Sentences

• Syntactic form indicates the way the words are
  related to each other in a sentence
• Logical form identifying the semantic
  relationships between words based solely on the
  knowledge of the language (independently of the
  situation)
• Final meaning representation mapping the
  information from the syntactic and logical form
  into knowledge representation
• System uses knowledge representation to represent
  and reason about its application domain

10
Parsing a Sentence

• Parsing determining the structure of the
  sentence according to the grammar
• Tree representation of a sentence
• Transition network grammars
• Start with initial node
• Can traverse an arc only if it is labeled with an
  appropriate category

11
Stochastic Methods for Ambiguity Resolution I

• Some sentences can be parsed many different
  ways, e.g. time flies like an arrow
• The most popular method for this is based on
  statistics
• Some facts from probability theory
• The concept of the random variable, e.g. the
  lexical category of flies
• Probability function
• assigns probability to every possible value of
  the random variable, e.g. 0.3 for flies being a
  noun, 0.7 for its being a verb
• conditional probability functions (Pr(AB)), e.g.
  the probability for the occurrence of a verb
  given the fact that a noun already occurred

12
Stochastic Methods for Ambiguity Resolution II

• Probabilities are used to predict future events
  given some data about the past
• Maximum likelihood estimator (MLE)
• Probability of X happening in the future number
  of cases of X happening in the past/total number
  of events in the past
• Works well only if X occurred often, not very
  useful for low-frequency events
• Expected likelihood estimator (ELE)
• Probability of X happening in the future
  f(number of cases of X happening in the
  past)/Sum(f(number of cases of some event
  happening in the past)), e.g. if
  f(Pr(X))Pr(X)0.5 and we know that Pr(X)0.4 and
  Pr(Y)0.6, then ELE(X)(0.40.5)/(0.40.50.60.5)
  0.45
• MLE is a special case of ELE, i.e. for MLE
  f(Pr(X)Pr(X)
• Given a large amount of text, one can use MLE or
  ELE to determine the lexical category of an
  ambiguous word, e.g. the word flies

13
Stochastic Methods for Ambiguity Resolution III

• Always choosing the interpretation that occurs
  most frequently in the training set on average
  obtains 90 success rate (not good)
• Some of the local context should be used to
  determine the lexical category of a word
• Ideally, for a sequence of words w1,w2,,wn we
  want a lexical category sequence c1,c2,,cn which
  maximizes the probability of right interpretation
• In practice, approximations of such probabilities
  are made

14
Stochastic Methods for Ambiguity Resolution IV

• n-gram models
• Look at the probability of a lexical category Ci
  which follows the sequence of lexical categories
  Ci-1,Ci-2,,Ci-n1
• Probability of c1,c2,,ck occurring is
  approximately the product of n-gram probabilities
  for each word, e.g. the probability of a sequence
  ART, N, V is 0.7110.43.3053
• In practice, bigram or trigram models are used
  most often
• The models capturing the concept are called
  Hidden Markov Models

15
Stochastic Methods for Ambiguity Resolution V

• In order to determine the most likely
  interpretation of a given sequence of n words, we
  want to maximize the value of
• The Viterbi algorithm
• Given k lexical categories, the total number of
  possibilities to consider for a sequence of n
  words is kn
• The Viterbi algorithm reduces this number to
  constnk2

16
Logical Form

• Although interpreting sentence often requires the
  knowledge of the context, some interpretation can
  be done independently of it
• basic semantic properties of a word, its
  different senses etc.
• Ontology
• each word has 1 or more senses in which it can be
  used, e.g. go has about 40 senses
• the different senses of all the words of a
  natural language are organized into classes of
  objects, such as events, actions etc.
• the set of such classes is called an ontology
• Logical form of an utterance can be viewed as a
  function that maps current discourse situation
  into a new one resulting from the occurrence of
  the utterance

17
Current Situation

• Inexpensive software for speech recognition
• The issues large vocabulary, continuous speech
  and speaker independence
• Automated speech recognition for restricted
  domains
• The speed of serial processes in a computer vs.
  the number of parallel processes in human brain

18
References

• Survey of the State of the Art in Human Language
  Technology, edited by Ronald A. Cole, 1996
• James Allen. Natural Language Understanding, 1995
• Raymond Kurzweil. When will HAL understand what
  we are saying? Computer Speech Recognition and
  Understanding. Taken from HALs Legacy, 1996