CompUXLSA: A computational model in predicting user experience from reviews using Latent Semantic Analysis (Conference Paper)
In this paper, we proposed a novel method (CompUXLSA) to predict user experience from reviews sentences using Latent Semantic Analysis (LSA). Human uses words to represent or express thoughts. The "word of mouth" could influence others especially through web and social media, which are the common communication tools today. We believe that reviews can be categorized according to user experiences since reviews are the thoughts and opinions from users after they have used certain products. In our works, we intend to mine and predict the user experience of expressed through reviews according to the five behavioral variables: Perceived Ease of Use, Perceived Usefulness, Affects towards Technology, Social Influence and Trust. We apply the state of the art method: Latent Semantic Analysis to build a semantic space and map review sentences to the most similar variable measurement items that adapted from Human Behavior Project to predict their experiences. Besides that, we also proposed a rule based template, SubEx to extract features of subject-experience from reviews to enhance the performance. Based on the results obtained, CompUXLSA had achieved average F-measure of 0.24. © 2015 IEEE.
Latent Semantic Analysis; opinion mining; user experience
Engineering controlled terms: Behavioral research; Control systems; Economic and social effects; Forecasting
Communication tools; Computational model; Latent Semantic Analysis; Opinion mining; Perceived ease of use; Perceived usefulness; State-of-the-art methods; User experience
Engineering main heading: Semantics
© Copyright 2016 Elsevier B.V., All rights reserved.