VIPE: A new interactive classification framework for large sets of short texts - application to opinion mining

TL;DR

VIPE is an interactive classification framework that efficiently labels large sets of short texts, such as tweets or opinion polls, by combining manual pre-classification with fast matrix factorization and user feedback.

Contribution

It introduces a novel interactive opinion mining tool that integrates user corrections in real-time using a fast matrix factorization approach for large short text datasets.

Findings

Effective classification of large short text datasets.

High-quality results confirmed by experiments and user feedback.

Suitable for real-time interactive opinion mining applications.

Abstract

This paper presents a new interactive opinion mining tool that helps users to classify large sets of short texts originated from Web opinion polls, technical forums or Twitter. From a manual multi-label pre-classification of a very limited text subset, a learning algorithm predicts the labels of the remaining texts of the corpus and the texts most likely associated to a selected label. Using a fast matrix factorization, the algorithm is able to handle large corpora and is well-adapted to interactivity by integrating the corrections proposed by the users on the fly. Experimental results on classical datasets of various sizes and feedbacks of users from marketing services of the telecommunication company Orange confirm the quality of the obtained results.