Multiple Instance Learning Networks for Fine-Grained Sentiment Analysis

Abstract

We consider the task of fine-grained sentiment analysis from the perspective of multiple instance learning (MIL). Our neural model is trained on document sentiment labels, and learns to predict the sentiment of text segments, i.e. sentences or elementary discourse units (EDUs), without segment-level supervision. We introduce an attention-based polarity scoring method for identifying positive and negative text snippets and a new dataset which we call SPOT (as shorthand for Segment-level POlariTy annotations) for evaluating MIL-style sentiment models like ours. Experimental results demonstrate superior performance against multiple baselines, whereas a judgement elicitation study shows that EDU-level opinion extraction produces more informative summaries than sentence-based alternatives.

Final published version. Please cite using appropriate date (2018). Link to journal: http://www.transacl.org/ojs/index.php/tacl/article/view/1225/277

References in corpus (6)

• ADADELTA: An Adaptive Learning Rate Method
• SummaRuNNer: A Recurrent Neural Network based Sequence Model for Extractive Summarization of Documents
• Seeing stars: Exploiting class relationships for sentiment categorization with respect to rating scales
• Semi-supervised Convolutional Neural Networks for Text Categorization via Region Embedding
• Neural Summarization by Extracting Sentences and Words
• Better Document-level Sentiment Analysis from RST Discourse Parsing