A multimedia interface for facilitating comparisons of opinions

Abstract

Written opinion on products and other entities can be important to consumers and researchers, but expensive and difficult to analyze. We present a multimedia interface designed to facilitate the analysis of opinions on multiple entities, which could be beneficial to many individuals and organizations. It integrates an information visualization and an intelligent system that selects notable comparisons in the data and summarizes them in text. This system applies a set of statistics for comparing opinions across entities. We conducted a study of our interface with 36 subjects. Subjects liked the visualization overall and our system's selections overlapped with those of subjects more than did the selections of baseline systems. Given the choice, subjects sometimes changed their selections to be more consistent with those of our system. This suggests that system selections were valuable to them.

OpinionBlocks: A Crowd-Powered, Self-improving Interactive Visual Analytic System for Understanding Opinion Text

“…Yatani et al developed Review Spotlight, which presents a word-cloud summary of online reviews in noun-adjective pairs [31]. Carenini and Rizoli built a multimedia interface that facilitates the comparison of different reviews [7]. More recently, Huang et al presented RevMiner, an interactive system that summarizes reviews in noun-adjective pairs to be presented in a compact mobile phone interface [13]; and Rohrdantz et al designed a visualization system that supports feature-based sentiment analysis of time-stamped review documents [26].…”

“…Not only do product reviews provide great value to individual consumers and influence their purchasing decisions [24], but they also impact the product or service strategies of businesses [32]. However, gaining insights becomes increasingly challenging for users as the number of reviews gets larger and larger [7,12,13,31].…”

“…While a few high-quality reviews or the aggregated sentiment may provide useful information, previous research shows that users often desire finer-grained understanding of the reviews [7,31]. In particular, people process information in an attributedriven manner in the absence of actual products (e.g., online shopping) [15].…”

“…To facilitate attribute-driven evaluation of products, a number of systems produce an aspect-based summary, including the extraction of sentiment toward each of the aspects [7,13,16,31]. Among these systems, several recent ones use noun-adjective pairs to summarize the aspects of a product/service (noun) and the sentiment (adjective) toward each aspect [13,31].…”

Speech for Content Creation

“…A follow-up study showed that these refinements to the options space, when volunteered by a system, help reduce dialogue turns [47]. Carenini and Rizoli specifically examine the presentation of opinion-based data in a multimedia setting and argue for the inclusion of dissimilar information and data on the degree of "controversiality" of opinions (i.e., how split the opinions were between positive and negative) [15].…”