Putting words in context: LSTM language models and lexical ambiguity
This addresses the problem of lexical ambiguity in language models for NLP researchers, but it is incremental as it focuses on analyzing an existing model rather than proposing a new solution.
The study investigated how an LSTM language model handles lexical ambiguity in English by probing its hidden representations for lexical and contextual information, finding that both are represented to a large extent but contextual information has room for improvement.
In neural network models of language, words are commonly represented using context-invariant representations (word embeddings) which are then put in context in the hidden layers. Since words are often ambiguous, representing the contextually relevant information is not trivial. We investigate how an LSTM language model deals with lexical ambiguity in English, designing a method to probe its hidden representations for lexical and contextual information about words. We find that both types of information are represented to a large extent, but also that there is room for improvement for contextual information.