Text analytics plays an important role in understanding users' preferences or focus of attention through the analysis of text written by users, especially in the area of e-commerce. A typical application of text analytics is the use of sentiment polarity detection to discover customer opinion on merchants or goods from comments or reviews made by customers. AAi data scientist and senior lecturer Dr Guandong Xu adopts text mining in recommender systems to improve the accuracy of predicting which merchants users are most likely to visit. His work provides an aggregated view by considering both individual preferences and community interest via utility theory, in relation to which, Dr Xu has developed an online demo system.
Reviews and the expression of opinions are prevalent in today’s online social networks. Most are in text format, thus by analysing this text, scientists are able to understand users' preferences and concerns. Moreover, this textual data provides an additional means for merchants to improve their services by addressing significant concerns raised by customers. As such, text analytics undoubtedly has an important role to play in the process of successfully conducting business.
AAi data scientist and senior lecture Dr Guandong Xu and his team are investigating the embedding of text analytics in recommender systems. By gaining better knowledge of individual preferences through text mining, recommender systems will become increasingly robust, as well as being capable of explaining the reason for specific system recommendations to customers. By defining several key aspects through text analytics, user preference and merchant quality can be precisely determined. Matching user preference with merchant quality leads to optimal recommendation and consequently encourages customer loyalty.
Dr Xu implemented the model and algorithm in a demo recommender system by utilizing real-world Yelp data via a mobile application. Unlike conventional recommender systems, which only provide the user with a selection list , the demo system gives the reason for specific recommendations. This work is to be presented at the top international conference on data engineering, ICDE 2015, and in the journal Expert Systems with Applications (ESWA). Dr. Guandong Xu can be reached at guandong.xu@uts.edu.au

Dr Guandong Xu delivers Text Analytics workshop