Now days, online buyer are so much aware and sensitive to product reviews. Usually before buying any product from e-commerce website they use to read products reviews and ratings. That’s why it is too much necessary for e-commerce website owners to keep watch on product reviews and its description. Users use to blame e-commerce websites if they sell products with bad reviews rather than products manufacturers which may ruin the reputation of e-commerce website brand. Sometimes competitators use to give fake reviews to improve their sells. Hence, it becomes too important for e-commerce website owner to detect fake product reviews and remove it from portal by doing proper sentimental analysis. For such e-commerce website owners, we have created one product system which is “Fake Product Review Detection and Removal System with Sentimental Analysis”.
In order to find-out fake reviews from e-commerce portal, we have developed product review monitoring cum removal system with proper sentimental analysis of genuine reviews framework. This genuine review framework can able to detect fake surveys taken via distinguished IP addresses by social media optimization teams. In this system, we have provided e-commerce owner login id for accessing framework with one secured key. Through that Id, owner can overlook on various items and can give survey about those items. To find out the audit is certified, framework will track IP address of the client if the framework watch counterfeit survey send by a similar IP Address numerous multiple times it will illuminate the administrator to expel that survey from the framework. This system helps the user to find out correct review of the product and remove it from portal if it is fake.
Following section represents Sentimental Analysis process to identify products fake reviews and how to remove them from portal.
Step 1: Data Preprocessing: Here, we have processed e-commerce portal product review data using following steps:
- First of all we start formatting data to represent it in a proper format for ML
- In second step, we clean up data to remove incomplete variables
- At last, we sample data and reduce it in run time for algorithms and memory requirements.
Step 2: Tokenization: In this step, we usually break the data into words, phrases and meaningful elements in order to explore the words presents in a sentence.
Step 3: Stop-word Elimination: Here , we mainly focused on text mining to identify negative stop words presents in reviews which should not be part of opinion and remove them from reviews.
Step 4: Bag-of-words Model: Here, we have to process our data for NLP and we only take here individual words into account to allot them specific subjectivity score.
Step 5: Training the classifier: Here, in this section we train our system for identification of fake product reviews by using predictive based test data analysis.
Step 6: Sentimental Analysis: For doing sentimental analysis of fake product reviews from database, here we take the use of Decision Tree Classifier and Naive Bayes and comparing the results.
In this article, we illustrated how with the help of proper sentimental analysis, we can identify fake product reviews and can able to remove it from our portal.