scholarly journals ECNU: Multi-level Sentiment Analysis on Twitter Using Traditional Linguistic Features and Word Embedding Features

Author(s):  
Zhihua Zhang ◽  
Guoshun Wu ◽  
Man Lan
Author(s):  
Aleksandra A. Talanina ◽  

Functional and stylistic studies give us an idea of linguistic features of speech products, thus enabling style identification. These specific features become most recognizable when comparing styles. Discourse studies, on the contrary, are mainly focused on understanding and describing basic factors of creating a form of a literary language (style) and factors that determine the characteristics of speech products in individual situations within a socially significant sphere. This article presents an analysis of the logical and compositional organization of the lecture as a genre of academic discourse, taking a university lecture from M. Mamardashvili’s course on M. Proust as an example. The specific nature of the lecture genre in academic discourse is determined by its basic function in the teaching process implemented in direct dialogue with the audience. The research is based on the thesis that a lecture is an event that can be analysed using the concept of chronotope. The use of this concept beyond the analysis of fiction is relevant since spatiotemporal coordination is mandatory for any speech product, regardless of the sphere it is created in or the functions it performs. The main feature of the lecture chronotope is multi-level organization, since a lecture has its own internal spatiotemporal coordinates. The lecture chronotope is explicated at different levels of the text (compositional, lexical and grammatical), which are interconnected. Considering this, two interconnected frameworks of the lecture – structural and semantic – are singled out; they provide the logical and compositional organization of the material, which is important to ensure students’ understanding.


Author(s):  
Pawan Kumar Verma ◽  
Prateek Agrawal ◽  
Ivone Amorim ◽  
Radu Prodan

2021 ◽  
pp. 199-211
Author(s):  
Bachchu Paul ◽  
Sanchita Guchhait ◽  
Tanushree Dey ◽  
Debashri Das Adhikary ◽  
Somnath Bera

2017 ◽  
Vol 44 (2) ◽  
pp. 184-202 ◽  
Author(s):  
Adel Assiri ◽  
Ahmed Emam ◽  
Hmood Al-Dossari

Sentiment analysis (SA) techniques are applied to assess aspects of language that are used to express feelings, evaluations and opinions in areas such as customer sentiment extraction. Most studies have focused on SA techniques for widely used languages such as English, but less attention has been paid to Arabic, particularly the Saudi dialect. Most Arabic SA studies have built systems using supervised approaches that are domain dependent; hence, they achieve low performance when applied to a new domain different from the learning domain, and they require manually labelled training data, which are usually difficult to obtain. In this article, we propose a novel lexicon-based algorithm for Saudi dialect SA that features domain independence. We created an annotated Saudi dialect dataset and built a large-scale lexicon for the Saudi dialect. Then, we developed our weighted lexicon-based algorithm. The proposed algorithm mines the associations between polarity and non-polarity words for the dataset and then weights these words based on their associations. During algorithm development, we also proposed novel rules for handling some linguistic features such as negation and supplication. Several experiments were performed to evaluate the performance of the proposed algorithm.


2021 ◽  
Vol 3 (2) ◽  
pp. 233-242
Author(s):  
Abdul Rahman Wahid Rapsanjani ◽  
Erfian Junianto

Penelitian ini bertujuan melakukan implementasi Probabilistic neural network dan Word Embedding dalam kasus sentiment analysis tentang tanggapan masyarakat tentang pemberian vaksin sinovac yangg diunggah di Twitter dan 3 class:positif, negative dan netral. Metode yang dipilih adalah metode klasifikasi Probabilistic Neural Network. Sebelum melakukan klasifikasi, praprocessing pada penelitian ini meliputi tokenizasi, normalisasi, menghilangkan emoticon, Convert Negasi, Stemming, Stopword Removal serta Word embedding. dataset yang digunakan berjumlah 1177 dataset dengan pembagiannya yaitu 560 dataset positif, 355 dataset negative dan 262 dataset netral. Program dirancang menggunakan Bahasa pemrograman python dengan beberapa library seperti keras, tensorflow dan pandas. Akurasi yang didapatkan pada pelatihan menggunakan Probabilistic  Neural Network sebesar 91%. Hasil pengujian adalah penelitian ini mampu melakukan sentiment analysis dengan kesalahan sebesar 9%.


Sign in / Sign up

Export Citation Format

Share Document