DSAS - PhD Public Lecture - Guandong Zhang

Title: How to Rank Answers in Text Mining

Abstract:

In our world, information is delivered and received through various data forms such as sound tracks, video clips, texts, and so on. Among all the data forms, texts are traditional and efficient way to deliver information. Among data analysis methods, we develop four methodologies to rank texts among a pool of collection. CEW-DTW and KL-CEW-DTW use time series to rank texts with respect to an “ideal” answer. Then, a general entropy is developed to find information in texts with respect to chosen keywords and group average. Finally, a Markov transition probability entropy is used to analyze not only information but also inner connections of keywords. Their performance will be demonstrated through a specific set of Amazon question/answer data.