Amazon review system
like u all know success of a company or product directly depends on its customer ,right ?so customer likes your product it’s your success if not then you certainly need to improvise it by making some changes in it so how will you know whether your product is successful or not ? well for that you need to analyze your customers and one of the attributes of analyzing your customers is to analyze the review of them and this is where the sentiment analysis comes into picture so starting with word sentiment analysis well you can define sentiment as process of computationally identifying and categorizing opinions from a piece of text and determine whether the writers attitude towards a particular topic or product ,is positive ,negative, or neutral well it might be possible not as an individual every time you don't perform a sentiment analysis but you look for feedback right? or people trust reviews .Amazon customers make sure to check online reviews of a product before they hit the buy button. A high no. of reviews suggest that the product is purchased by a large number of buyers while lots of positive reviews indicate that the product is of high quality. you can look for feedback like what the other customers have to say about that particular product weather it isn't good or bad right and you analyze it manually or by traditional method by looking at their feedbacks now consider at the company level so how will the company , what the customers is thinking about their product they don't have just one or two customers ? Right , they have more than millions of customers and as well as the data that you collect via review sites and social channels are all in unstructured format, which is difficult to analyze & the product based company like amazon needs to perform analysis so this is where Natural Language Processing and Machine Learning is so useful.
Lets move and see how does it actually work so lets take the statements as an example of amazon product reviews on jeans .” design of jeans is beautiful !”
step 1:This step include dividing a statement into different set of words ,so the statement ‘design of jeans is beautiful !’ would be further segregated into its different words i.e.,
design ,of ,jeans ,is ,beautiful,!
step 2:[cleaning the data] it means to remove all the special characters or any other word which do not add any value analytics part so as in step two it removes the special characters over here i.e. exclamation mark so it left with 5 words i.e.
design ,of ,jeans ,is ,beautiful
step 3: next step would be the removing the stop words ,so the stop words like the ,was ,is ,he ,she which do not add much value to the analytics part so we can easily remove them so now we are just left with 3 words i.e.
design ,jeans ,beautiful
step 4:[classification] Now we are left with only 3 words then your task is to classify them as weather it is +ve word or -ve word or neutral word. For +ve we give sentiment score as +1 ,for -ve we have -1 and for neutral the score is 0.Now this is the part where machine learning comes you can model /Train your data with dictionary of pre-classified set of word (lexicons) & now the once model is trained ,you can train perform the test /analyze statement and if more will be the accuracy score i.e. will be the classification .Right ? so well we classified our words as +ve & -ve words. Therefore the sentiment score of design and jeans is 0 as its Neutral & beautiful word is +1 i.e. +ve.
step 5: so lets move an ahead and calculate the final sentiment score of the whole sentence .now since we have just 3 words over here which adds some values to the analytics part .overall score by combining the statement , we have +1+0+0=+1 , so now you can say that since the polarity is >0 so the statement is +ve.
SO THE POLARITY IS >0 ,SO THE STATEMENT IS +VE
similarly, if we take the other example of reviews then they will also analyzed or brand through this steps ,so lets take the few example of reviews and there final labeling will be as given as below.
conclusion: sentiment analysis like in amazon is an ongoing process to understand what consumers think about your product or brand. maximum possible product reviews are collected from different sources sites ,social media platforms .with the increasing market competition ,sentiment analysis become the need of the time. even the established brand are actively using this technique to improve the consumer experience and company never recommend such product on their front page ,whos review is bad ,no matter if you have a new product or popular one ,you must adopt sentiment analysis in the ways mentioned above to constantly improve the user experience and stay ahead of the competitors