We would anallyz, oops analyze, a Small Text Data with the help of a Self-Service Tool. Please click on the link above to use the tool.
Please download the Sample File from the link. It has User reviews about a Phone. The reviews are identified with an ID, Country and Date. We would perform three simple exercises on this data. By doing that, we would utilize all the preliminary steps of Text Mining. We would remove redundant words, numbers, punctuation, convert the entire text to lower case, identify words with their Parts-of-speech etc.
We can see analogous applications in analyzing Customer Feedback, Employee Survey, Social Media Post etc.
Worcloud can be generated for the entire corpus of Text or can be Segregated based on Context too. As an example, we can find most frequent NOUN, VERB and ADJECTIVE separately. What does it tell us? Maybe "apps" (Noun) are the most talked about things, which user "like" (Verb) a lot, due to its "great" (adjective) features. Combination of the most frequent Nouns, Adjectives and Verb can hint at the overall theme of Text Data.
Then one can make a query on the specific reviews by one or combination of most frequent words.
Sentiment Analysis is done for each Dimension (Topic) as well as for each Comment. Hence the Sentiment Score for "Battery" based on all the Comments, and Sentiment Score for a user on his complete Comment (on Battery, App, Price etc.) is also available. The most Critical Dimension could be one, which evokes extreme Sentiments (maximum Positive as well as Negative) and has been commented by many users.
If we have location and time available for each reviews, we can look at the sentiment for each Country and its trend with time as well.
Download the Data, open the Tool and play with the Words, and relate to some of the concepts we learnt earlier.
Please watch the Youtube Video, which showcases the power of Text Mining. Guest reviews on a London Hotel is scraped, loaded in ANALLYZ and results are there to see. Wouldn't any Hotel, Airline or any Customer-oriented Business would love to do.