Aspect and sentiment unification model for online review analysis

By Yohan Jo, Alice H. Oh. Web Search and Data Mining, 2011

813 citations recorded by the source corpus.

User-generated reviews on the Web contain sentiments about detailed aspects of products and services. However, most of the reviews are plain text and thus require much effort to obtain information about relevant details. In this paper, we tackle the problem of automatically discovering what aspects are evaluated in reviews and how sentiments for different aspects are expressed. We first propose Sentence-LDA (SLDA), a probabilistic generative model that assumes all words in a single sentence are generated from one aspect. We then extend SLDA to Aspect and Sentiment Unification Model (ASUM), which incorporates aspect and sentiment together to model sentiments toward different aspects. ASUM discovers pairs of {aspect, sentiment} which we call senti-aspects. We applied SLDA and ASUM to reviews of electronic devices and restaurants. The results show that the aspects discovered by SLDA match evaluative details of the reviews, and the senti-aspects found by ASUM capture important aspects that are closely coupled with a sentiment. The results of sentiment classification show that ASUM outperforms other generative models and comes close to supervised classification methods. One important advantage of ASUM is that it does not require any sentiment labels of the reviews, which are often expensive to obtain.

Subfields: Artificial Intelligence

View Paper: https://www.semanticscholar.org/paper/c311778adb9484c86250e915aecd9714f4206050

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

Citations in the corpus

Survey on Aspect-Level Sentiment Analysis (2016, background): “same idea is demonstrated in both [21] and [74].”
Analysis of user comments: An approach for software requirements evolution (2013, methodology): “To address the joint problem of discovering the topics in reviews and the sentiments toward those topics, recent work has applied a combination of topic modeling and sentiment analysis to reviews of electronic devices and restaurants [15] as well as movies [16].”
Hidden factors and hidden topics: understanding rating dimensions with review text (2013, background): “We brieﬂy mention works where aspects are explicit , i.e., where users rate individual features of a product in addition to their overall rating [1,9,17,24], though this work differs from ours in that it requires multiple ratings per review.”
Explainable recommendation with fusion of aspect information (2018, methodology): “Sentiment-based methods utilize the sentiment analysis [2, 13, 20, 26] for recommendation.”
A Survey of Opinion Mining and Sentiment Analysis (2012, background): “Jo and Oh [41] proposed an Aspect and Sentiment Unification Model (ASUM) to model sentiments toward different aspects.”
Topic Modeling based Sentiment Analysis on Social Media for Stock Market Prediction (2015, methodology): “ASUM is a model for extracting both the aspect and sentiment for online product review dataset (Jo and Oh, 2011).”
Learning Attitudes and Attributes from Multi-aspect Reviews (2012, background): “Finally, [17] examine whether similar models can be applied without explicit aspect ratings, so that ratable aspects might be uncovered using only overall ratings.”
Mining topics in documents: standing on the shoulders of big data (2014, methodology): “Topic models have been used for the task by many researchers [5, 10, 16, 21, 23, 25, 26, 32, 33, 37].”
Improving Sentiment Analysis with Document-Level Semantic Relationships from Rhetoric Discourse Structures (2017, background): “Other approaches utilize low-level discourse-based features in the form of connectors for (sub-)clauses [7]–[9].”
Mining Aspect-Specific Opinion using a Holistic Lifelong Topic Model (2016, methodology): “sions), they are not specialized models nor fine-grained models for opinion mining as [25, 19, 52].”
FLAME: A Probabilistic Model Combining Aspect Based Opinion Mining and Collaborative Filtering (2015, methodology): “D-LDA is a state-of-theart aspect-based opinion mining model which can be seen as a generalization of several other models [20, 6].”
Current State of Text Sentiment Analysis from Opinion to Emotion Mining (2017, methodology): “For instance, one of the most important group of works in this category is devoted to utilizing topic modeling in aspect extraction such as the work of Lin and He [2009], Jo and Oh [2011], Mukherjee and Liu [2012], and Wang et al. [2016].”
ReEL: Review Aware Explanation of Location Recommendation (2018, background): “The problem of aspect extraction from review text has been quite popular [8, 28, 52] for various problems (e.g., rating prediction [31], aspect-sentiment summarization [24, 34, 42], recommendation [30, 53], etc.).”
Mining Marketing Meaning from Online Chatter: Strategic Brand Analysis of Big Data Using Latent Dirichlet Allocation (2014, methodology): “To implement this, we use an initial small set of seed words that are unambiguously positive (e.g., good, great) or negative (e.g., bad, horrible, lousy) irrespective of the market, context, or dimension (Jo and Oh 2011; Lin and He 2009; Turney and Littman 2003).”
Description Based Text Classification with Reinforcement Learning (2020, background): “…such as sentiment analysis (Pang et al., 2002; Maas et al., 2011; Socher et al., 2013; Tang et al., 2014, 2015b), aspect sentiment clas-siﬁcation ( Jo and Oh, 2011; Tang et al., 2015a; Wang et al., 2015; Nguyen and Shirai, 2015; Tang et al., 2016b; Pontiki et al., 2016; Sun et al., 2019a), topic…”