An extractive text summarization based on multivariate approach

The work presents a method to automatically summarize a text with the help of multivariate statistical technique. The method is subdivided into two phases namely the training and the testing phases. The goal of this approach is to design a summarizer which will perform the task very much like a human. We propose a training methodology where we train the system using manual summaries. The utilization of multivariate statistical technique for this task is justified by its ability to produce a model that resembles a relation. Primary subjective evaluation shows that our approach is effective, efficient and the performance of the system is promising.