Today's marketing and research world social media monitoring, brand monitoring, the voice of the customer, customer service, and market research become a necessity. And marketers and researchers feel that determining how the target audience feels and thinks about a product or service is a very tough and painful process. Additionally, manual analysis becomes even more complicated when the size of the data collected is consistently increasing, and large data sets come into play. Just then, the English Sentiment Classification Model comes to the rescue for understanding practically what sentiment a text contains. With this English Sentiment Classification Model, you can easily classify your data as "positive" or "negative" (no matter how many rows data has) and get the results of your work in the fastest way. So, you can see where the problem is and have more time to take any action at the end of the day.
About Training Set
The dataset was prepared by collecting social media data from different industries and topics. The data came from the Analytics platform of Kimola and labeled as "positive" or "negative". It was paid attention to whether a sentence contains a "positive" or "negative" feeling while preparing the training set. Also, it was crucial to divide the sentences with opposite sentiment and remove proper names such as brand names, company names, and celebrity names. Training data was trained three times using cross-validation with 3,305 rows of data, and this set is still under development.