Cword2vec: a novel morphological rule-based word embedding approach for Urdu text sentiment analysis

Area of Science:

• Natural Language Processing (NLP)
• Computational Linguistics
• Machine Learning

Background:

• Word embeddings are crucial for NLP, capturing syntactic and semantic word information.
• Urdu, spoken by over 231 million people, lacks sufficient NLP research, especially concerning its complex word structures.
• Urdu's unspecified word boundaries and prevalence of compound words pose challenges for traditional segmentation methods like bigrams and trigrams.

Purpose of the Study:

• To address the challenges of Urdu compound words in NLP.
• To propose a novel morphological rule-based compound word embedding (Cword2vec) for Urdu text representation.
• To evaluate the effectiveness of Cword2vec in Urdu sentiment analysis using deep learning models.

Main Methods:

• Developed a self-trained morphological rule-based compound word embedding (Cword2vec) model based on word2vec.
• Applied Cword2vec for text representation in Urdu sentiment analysis tasks.
• Evaluated Cword2vec performance using Long Short-Term Memory (LSTM), Bidirectional LSTM (BiLSTM), Convolutional Neural Networks (CNN), and Convolutional LSTM (C-LSTM).
• Compared Cword2vec against traditional bigram and trigram approaches for compound word identification.

Main Results:

• The proposed Cword2vec model demonstrated superior performance across all evaluated deep learning models.
• Morphological rule-based compound word embeddings significantly outperformed traditional bigram and trigram methods.
• Improvements were observed in key metrics including precision, recall, F1 score, and accuracy.

Conclusions:

• Morphological rule-based compound word embeddings offer a more effective approach for Urdu NLP tasks compared to traditional methods.
• Cword2vec provides a robust solution for handling Urdu's complex word structures, enhancing sentiment analysis.
• Further research into Urdu NLP, particularly with morphological insights, is warranted to leverage its linguistic richness.