An Unsupervised Neural Attention Model for Aspect Extraction

Abstract

Aspect extraction is an important and challenging task in aspect-based sentiment analysis. Existing works tend to apply variants of topic models on this task. While fairly successful, these methods usually do not produce highly coherent aspects. In this paper, we present a novel neural approach with the aim of discovering coherent aspects. The model improves coherence by exploiting the distribution of word co-occurrences through the use of neural word embeddings. Unlike topic models which typically assume independently generated words, word embedding models encourage words that appear in similar contexts to be located close to each other in the embedding space. In addition, we use an attention mechanism to de-emphasize irrelevant words during training, further improving the coherence of aspects. Experimental results on real-life datasets demonstrate that our approach discovers more meaningful and coherent aspects, and substantially outperforms baseline methods on several evaluation tasks.

A Unified Model for Opinion Target Extraction and Target Sentiment Prediction

“…Traditionally, this task can be broken into two sub-tasks, namely, opinion target extraction and target sentiment classification. The goal of opinion target extraction is to detect the opinion target mentions in the text, and it has been extensively studied (Qiu et al 2011;Liu, Xu, and Zhao 2013;Liu, Xu, and Zhao 2014;Liu, Joty, and Meng 2015;Yin et al 2016;Wang et al 2016a;He et al 2017;Li and Lam 2017;Li et al 2018b;. The second sub-task, i.e., target sentiment classification, performs as a multiplier for the usefulness of the extracted target mentions, as it can predict the sentiment polarity of the given opinion targets.…”

Exploiting Document Knowledge for Aspect-level Sentiment Classification

“…where t is the target representation computed as the averaged word embedding of the target. f score is a content-based function that captures the semantic association between a word and the target, for which we adopt the formulation used in (Luong et al, 2015b;He et al, 2017) with parameter matrix W a ∈ R d×d . The sentence representation z is fed into an output layer to predict the probability distribution p over sentiment labels on the target:…”

AspeRa: Aspect-Based Rating Prediction Model

“…Since this model utilizes an aspect embedding matrix to approximate aspect words in the vocabulary, initialization of aspect embeddings is crucial. The work [8] used k-means clustering-based initialization [17,18,36], where the aspect embedding matrix is initialized with centroids of the resulting clusters of word embeddings. We compare two word embeddings for AspeRa: GloVe [29] and word2vec [21,23].…”