Predicting event mentions based on a semantic analysis of microblogs for inter-region relationships

Abstract

An ability to predict people’s interests in different regions would be valuable to many applications including marketing and policymaking. We posit that social media plays an important role in capturing collective user interests in different regions and their dynamics over time and across regions. Event mentions in microblogs of social media like Twitter not only reflect the people’s interests in different regions but also affect the posting of future messages as the content of microblogs propagates to others through an online social network. Differentiating from the various network analysis techniques that have been developed to capture people’s interests and their propagation patterns, we propose an event mention prediction method that utilises an analysis of inter-region relationships. We first obtain regional user interests for each topic by applying Latent Dirichlet Allocation (LDA) to region-specific collections of tweets and then compute pairwise similarities among regions. The resulting similarity-based region network becomes the basis for constructing region groups through Markov Cluster Algorithm, which helps removing noise relationships among regions. We then propose a relatively simple regression technique to predict future event mentions in different regions. We demonstrate that the proposed method outperforms the state-of-the-art event prediction method, confirming that the novel method of constructing groups from region-based sub-topic interests indeed contributes to the increase in the prediction accuracy.

Twitter Sentiment Analysis Approaches: A Survey

“…Some researchers have focused on analyzing twitter data using SNA metrics and similarity-based algorithms. For instance, the work in [36] analyzed event mentions in microblogs of social media, like twitter, for quantify-ing user's interests using similarity-based region network. Regional user interests are obtained for each topic by applying latent Dirichlet allocation to region-specific collections of tweets, and then compute pairwise similarities among regions.…”

E-Commerce Personalized Recommendation Model Based on Semantic Sentiment

“…To put it simply, just like when Taobao is usually opened, the system will present the products you may need to buy on the main page without any input from the customer. Relevant studies have shown that after using the e-commerce personalized recommendation system, the sales of goods can be increased by 2% to 8% [10].…”

Time-aware domain-based social influence prediction

“…analysed event mentions in microblogs of social media, like Twitter, in order to quantify users' interests using a similarity-based regional network [38].…”