click here to download the abstract project
click here to download the base paper
ABSTRACT
Spam Checker Analysis systems are basically binary class or multi-class classification system that classifies feedback of customers into various Specific classes. It has become an industry on its own. There are dozens of notable internet companies, which can be referred as app companies who are doing customer feedback analysis for other, often much larger companies. Many feedback companies like Freshdesk and Nebula do analysis to various internet age companies like Amazon , Google ,Microsoft . Multiclass classification of texts is the challenging task in the field of Machine Learning. Our work focus at the task of six class classification of feedbacks received from different customers. Our corpus is categorized into six class classification namely comment, request, bug, complaint, meaningless and undetermined. The training and testing sets are generated using 10-fold Cross-validation method. We have achieved an accuracy of 64.70% using Random Forest algorithm. However, he baselines accuracy achieved by
us is 53.42% using Gaussian Naïve Bayes algorithm.
Introduction:
Spam Checker is a marketing term that describes the process of obtaining a spam opinion about a business, product or service. Spam checker is so important because it provides marketers and business owners with insight that they can use to improve their business, products and/or overall spam experience. Spam analysis measures how happy spam are with a company’s products and services. Feedback analysis provides companies with feedback about everything from products to the buying process to support. Most organizations combine this powerful data with other forms of spam to create actionable intelligence about the entire spam journey. Spam is the one thing that gives a business a clearer view of how it is doing. Proper analysis provides a business with a better view of what it has to change, what it has to improve on, and what it has to do, to retain and grow revenue and profit. Each customer review counts. But it is practically impossible to go through each customer manually. Hence a robust and an efficient system is necessary for the classification of the spam checker of a commodity. We aim to classify a feedback into six classes, namely comment, request, bug, complaint, meaningless and undetermined (further discussion has been done in chapter 4). Using past data we train our system to learn and hence use that acquired knowledge to assign a new spam checker into its suitable class. The whole process makes the evaluation of a product in the future market easy This thesis consists of 6 chapters.
chapter 1 introduces us to the problem statement while chapter 2 elaborates the related work done in this field. chapter 3 discusses the tools that we have used in the project. chapter 4 deals with the implementation of our work in detail. chapter 5 displays the observations obtained and consequently.
