Browsing by Author "AMINA AMJAD"

Now showing 1 - 1 of 1
  • No Thumbnail Available
    Item
    Text sentiment analysis for categorizing customer response
    (UMT.Lahore, 2019) AMINA AMJAD
    Social media has completely reformed the concentration of marketing from a trader to buyer perspective. Nowadays, flow of marketing is controlled by customers rather than companies because of excess of information on social media sites. The Web facilitates us with a virtual world where customers can experience products before buying. Customer behavior includes the intentions and the actions they perform in the consumption and usage of a product. Customer intentions, feelings and activities are constantly changing.In a recent era, the drastic increase in the usage of mobile phones and social media has become a way for knowing customer feedbacks on various platforms and regarding number of services or products. Analyzing the social feedbacks of the customer for business aspect brought a new thought towards sentiment classification. In order to inject accuracy in a prediction model require a wide and clear approach towards data analysis and interpreting user intention. Natural language processing is providing large scope of text analysis by developing opinion classification model. In this thesis we propose the sentiment classification model for customer reviews on unlocked mobile phones obtained from Amazon customer reviews datasets by using sentiment analysis that is based on polar opinions. Customer generated feedback is classified for analysis. From the previous experiments we have decided to apply NLP techniques for this purpose. The primary importance of this research is to establish a better relation between relevantly extracting features and its classification when it comes to text based datasets. We have performed comparative analysis in order to get to know user emotions through feature extraction and classification algorithms. Sentiment text classification and evaluation is experimented and evaluated in different perspective. Customer intentions are classified into positive, negative and neutral sentiments. Data for this purpose is consumed from Amazon customer review .Text is labeled as polarity sentiments through classification with the probable accuracy of 70-93% for positive or negative reviews of customers by using this model. Evaluation of the system is carried out extensively by performance measure techniques. Keywords: Sentiment analysis, Amazon customer reviews, polarity sentiments, comparative analysis.