Opinion analysis of the university library readers

---

Oukrich Jaouad

Sultane Moulay Slimane University, Morocco

: J Comput Eng Inf Technol

Abstract

---

The last years were known to have a big particular interest about the opinion analysis. However, the evaluation of this task was not satisfactory because of the rarity or of the non-availability of the corpuses annotated by various languages and data processing methods have changed from analytics (ETL, SQL, and Reporting) into Big Data (Hadoop, NoSQL). This article describes our contribution on the opinion fine detection of the university library readers’ in social media, mainly the treatment of Big Data, which is extracted from the Facebook. The approach adopted for the readers' opinion analysis (combination of the statistical and symbolic methods) will be carried out by using a Java Hadoop Framework. Firstly, we describe the construction of the vocabulary that allows characterizing a positive, neutral or a negative opinion applied to the study of readers’ satisfaction. Secondly, we introduce the application results by using machine-learning techniques for the opinion classification task.

Biography

---