Data-Driven Approaches and Machine Learning
While human language users make highly efficient use of the ambiguity in natural language, this presents a major challenge to computational NLP. A central factor when we try to determine the most likely interpretation of a given utterance is earlier experience (for example in the form of observed frequencies of word combinations and grammatical structures). Language Technology today models such phenomena in terms of probabilistic models of language structure, mathematically complex systems that ‘learn’ generalizations over usage patterns from a training distribution.
LTG members are interested in various approaches to such data-driven modeling for a range of NLP tasks, for example dependency and unification-based parsing, natural language generation, dialogue modeling, lexical acquisition, and sentiment analysis.