Beyond Word Importance: Contextual Decomposition to Extract Interactions from LSTMs


Nov 03, 2017 (modified: Nov 03, 2017) ICLR 2018 Conference Blind Submission readers: everyone Show Bibtex
  • Abstract: The driving force behind the recent success of LSTMs has been their ability to learn complex and non-linear relationships. Consequently, our inability to describe these relationships has led to LSTMs being characterized as black boxes. To this end, we introduce contextual decomposition (CD), a novel algorithm for capturing the contributions of combinations of words or variables in terms of CD scores. On the task of sentiment analysis with the Yelp and SST data sets, we show that CD is able to reliably identify words and phrases of contrasting sentiment, and how they are combined to yield the LSTM's final prediction. Using the phrase-level labels in SST, we also demonstrate that CD is able to successfully extract positive and negative negations from an LSTM, something which has not previously been done.
  • TL;DR: We introduce contextual decompositions, an interpretation algorithm for LSTMs capable of extracting word, phrase and interaction-level importance score
  • Keywords: interpretability, LSTM, natural language processing, sentiment analysis, interactions