So what are we discussing here?
Problem being addressed
Millions of online discussions are generated everyday on social media platforms. While topic modelling is an efficient way of better understanding large text datasets at scale, the conventional topic models have had limited success in online discussions due to sparsity of data and “noisy” comments, which are good for sentiment analysis but are significant disturbances when extracting topics from discussion threads.
A novel model that uses the discussion thread tree structure and propose a “popularity” metric to quantify the number of replies to a comment to extend the frequency of word occurrences. The suggested model is based on popularity and transitivity to infer topics and their assignments to comments.
Advantages of this solution
By comparing the proposed model with a number of state-of-the-art baseline models on real word datasets, the researchers have demonstrated competitive results, and the effectiveness of using conversational discourse structure to help in identifying topical content embedded in short and colloquial online discussions. Experiments on real forum datasets were used to demonstrate improved performance for topic extraction with six different measurements of coherence and impressive accuracy for topic assignments.
Solution originally applied in these industries
Possible New Application of the Work
The model can assist users to browse a long discussion thread quickly by summarizing the possible topics. Quite often a popular news article or interesting post can easily accumulate thousands of comments within a short period of time, which makes it difficult for interested users to access and digest information in such data. Similarly, students spend a lot of time looking for the relevant topic discussions on different types of sites, from Quora to Stackoverflow, and the model will make this search more efficient.
The suggested model is very helpful for sentiment analysis or stance detection. The massive amount of online discussions provides us with valuable resources for studying and understanding public opinions on fundamental societal issues, e.g., abortion or gun rights. Automatically predicting user stance and identifying corresponding arguments are important tasks for improving policy-making process and public deliberation.
Source URL: #############