Multiple Instance Learning Networks for Fine-Grained Sentiment Analysis (2017-11-27T00:00:00.000000Z)

TL;DR

An attention-based polarity scoring method for identifying positive and negative text snippets and a new dataset which is called SpoT (as shorthand for Segment-level POlariTy annotations) for evaluating MIL-style sentiment models like the authors' are introduced.

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.

Authors

Mirella Lapata

27 papers

S. Angelidis

6 papers

TL;DR

Abstract

Authors

References62 items

Explicit Document Modeling through Weighted Multiple-Instance Learning

SummaRuNNer: A Recurrent Neural Network Based Sequence Model for Extractive Summarization of Documents

Improving Multi-Document Summarization via Text Classification

The Role of Discourse Units in Near-Extractive Summarization

Hierarchical Attention Networks for Document Classification

Neural Summarization by Extracting Sentences and Words

Document Modeling with Gated Recurrent Neural Network for Sentiment Classification

Better Document-level Sentiment Analysis from RST Discourse Parsing

From Group to Individual Labels Using Deep Features

Deep multiple instance learning for image classification and auto-annotation

Semi-supervised Convolutional Neural Networks for Text Categorization via Region Embedding

Extraction of Salient Sentences from Labelled Documents

Scalable Multi-instance Learning

Effective Use of Word Order for Text Categorization with Convolutional Neural Networks

Extracting Lexically Divergent Paraphrases from Twitter

Abstractive Summarization of Product Reviews Using Discourse Structure

Explaining the Stars: Weighted Multiple-Instance Learning for Aspect-Based Sentiment Analysis

Neural Machine Translation by Jointly Learning to Align and Translate

Convolutional Neural Networks for Sentence Classification

Jointly modeling aspects, ratings and sentiments for movie recommendation (JMARS)

Confidence-Rated Multiple Instance Boosting for Object Detection

A Hybrid Approach to Multi-document Summarization of Opinions in Reviews

Distributed Representations of Sentences and Documents

Recursive Deep Models for Semantic Compositionality Over a Sentiment Treebank

ADADELTA: An Adaptive Learning Rate Method

Text-level Discourse Parsing with Rich Linguistic Features

Baselines and Bigrams: Simple, Good Sentiment and Topic Classification

Semi-Supervised Recursive Autoencoders for Predicting Sentiment Distributions

Knowledge-Based Weak Supervision for Information Extraction of Overlapping Relations

Lexicon-Based Methods for Sentiment Analysis

Discovering Fine-Grained Sentiment with Latent Variable Structured Prediction Models

Learning from Partial Labels

Opinosis: A Graph Based Approach to Abstractive Summarization of Highly Redundant Opinions

Exploring the Use of Word Relation Features for Sentiment Classification

The Bag-of-Opinions Method for Review Rating Prediction from Sparse Text Patterns

SentiWordNet 3.0: An Enhanced Lexical Resource for Sentiment Analysis and Opinion Mining

Sentiment Summarization: Evaluating and Learning User Preferences

Multi-instance learning by treating instances as non-I.I.D. samples

Learning to Recognize Objects with Little Supervision

Multiple Instance Boosting for Object Detection

Automatic Text Summarization of Newswire: Lessons Learned from the Document Understanding Conference

Seeing Stars: Exploiting Class Relationships for Sentiment Categorization with Respect to Rating Scales

Mining and summarizing customer reviews

Logistic Regression and Boosting for Labeled Bags of Instances

Multiple-Instance Learning via Disjunctive Programming Boosting

A Two-Level Learning Method for Generalized Multi-instance Problems

Content-Based Image Retrieval Using Multiple-Instance Learning

A Noisy-Channel Model for Document Compression

Thumbs Up or Thumbs Down? Semantic Orientation Applied to Unsupervised Classification of Reviews

Thumbs up? Sentiment Classification using Machine Learning Techniques

Building a Discourse-Tagged Corpus in the Framework of Rhetorical Structure Theory

Multiple-Instance Learning for Natural Scene Classification

Solving the Multiple Instance Problem with Axis-Parallel Rectangles

A Self-Organizing Integrated Segmentation and Recognition Neural Net

Book Review: Computer Intensive Methods for Testing Hypotheses: An Introduction

Rhetorical Structure Theory: Toward a functional theory of text organization

GICF: The Group-Instance Cost Function model introduced in Kotzias et al.

Fully-supervised CNN segment classi-ﬁer trained on S PO T’s labels

Multi-Document Summarization of Evaluative Text

Annotating Expressions of Opinions and Emotions in Language

on Computational Linguistics: Posters

Polarity source: Controls whether we assign polarities via segment-speciﬁc or document-wide predictions

Field of Study

Journal Information

Name

Page

Volume

Venue Information

Name

Type

URL

Alternate Names