An Embedding-based Joint Sentiment-Topic Model for Short Texts
This addresses the problem of understanding user behavior from short texts for industries like service and healthcare, though it appears incremental as it builds on existing joint sentiment-topic models.
The authors tackled the challenge of extracting coherent sentiment-associated topics from short texts by developing ELJST, an embedding-enhanced joint sentiment-topic model, which achieved average improvements of 10% in topic coherence and 5% in topic diversification over baselines.
Short text is a popular avenue of sharing feedback, opinions and reviews on social media, e-commerce platforms, etc. Many companies need to extract meaningful information (which may include thematic content as well as semantic polarity) out of such short texts to understand users' behaviour. However, obtaining high quality sentiment-associated and human interpretable themes still remains a challenge for short texts. In this paper we develop ELJST, an embedding enhanced generative joint sentiment-topic model that can discover more coherent and diverse topics from short texts. It uses Markov Random Field Regularizer that can be seen as a generalisation of skip-gram based models. Further, it can leverage higher-order semantic information appearing in word embedding, such as self-attention weights in graphical models. Our results show an average improvement of 10% in topic coherence and 5% in topic diversification over baselines. Finally, ELJST helps understand users' behaviour at more granular levels which can be explained. All these can bring significant values to the service and healthcare industries often dealing with customers.