Online consumer reviews have become an essential source of information for understanding markets and customer preferences. This research introduces a novel topic model to identify product attributes and sentiments toward them at the sentence level. The model uses a recursive definition of topic distribution in a sentence to avoid the problem of over-parametrization in topic models. The introduction of the inference network enables the utilization of rich features in the content to drive the identification of sentiments, in contrast with other multi-aspect sentiment analysis models that rely on single words. The sentence topic model has a superior performance in producing coherent topics, and the sentence topic-sentiment model outperforms the existing model on the task of predicting product attribute rating.
CITATION STYLE
Chen, T., & Parsons, J. (2018). A sentence-level sparse gamma topic model for sentiment analysis. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 10832 LNAI, pp. 316–321). Springer Verlag. https://doi.org/10.1007/978-3-319-89656-4_33
Mendeley helps you to discover research relevant for your work.