Sentiment Analysis of User Reviews about Hotel in Roman Urdu

In recent years, sentiment analysis has a significant role in various social media networks, electronic marketing websites, communication forums, and blogging websites. There are many issues in sentiment mining, classification, and analysis like huge lexicon, Natural language processing overhead, fake reviews, etc. Out of these issues, one major problem is that the comments and reviews can be in different languages like French, Chines, English, Urdu, Arabic, etc. To handle each language according to its syntax, semantic, and structure is a challenging task. Many researchers work on English, Urdu, and Arabic sentiment analysis, but very limited work has been done on resource constrain languages like Roman Urdu. In this paper, Python is used to execute different classification machine learning models for Roman Urdu text analysis. A total of 3000 reviews dataset has been scrapped from different hotel websites. The results show that logistic regression and SVM outperformed in terms of accuracy, recall, precision, and F-measure.