The problem Fashion Apps Analyzer solves

The application fetches Google Playstore Fashion App Reviews to conduct market research study on the behavioural analysis of customers and their frequency of categories preferred. It also does spam detection for the reviews and if a specific tag categorized as spam, then a ticket for spam is generated. The tags considered for categorization includes 'discounts_offers', 'app_interface', 'customer_support', 'ease_of_return', 'bug', 'feature_request', 'question' and 'feedback'.

Sentimental Analysis is included with each reviews depicting the frequency of choices of customers based on reviews. It shows what all are the keywords and tags customers are mostly using. To be specifc, what are the most reviews about and for each of the tag, sentimental analysis would be provided which shows, how many customers reviews are under each category. Customers describe about the discounts and offers on particular apps, and they can be identified using 'discounts_offers' tag. The 'app_interface' tag shows the overall interface of various apps.

The 'customer_support' tag provides analysis on how all the customers found the particular app to be customer friendly and provides enough support. The 'ease_of_return' tag analyse the reviews of customers regarding the return and replacement policies of a particular app. The reviews of customers on the bug of certain apps are depicted using 'bug' tag. 'feature_request' tag shows any request for features. The 'question' tag is for questions from customers. The 'feedback' tag categorizes the generalised feedback from customer reviews.

The above categorized, processed, spam detected and clustered reviews provide great insight for market research trends in fashion apps. If any startups or companies wanted to start any fashion apps, they can make use of this application to get better understanding and sentiments of the customers and hence provides more insights on current trends and improvements to make for the newer versions of apps.

Challenges we ran into

During the course of our project, we encountered several challenges, each requiring thoughtful solutions to overcome. One significant hurdle was integrating the Devrev snapin for accessing Google Play Store app reviews. This process involved navigating various technical requirements to ensure compatibility with our idea.

Additionally, implementing the clustering algorithm to categorize reviews into different tags posed its own set of challenges. We had to carefully design and fine-tune the clustering model to accurately identify patterns and group similar reviews together. This required extensive testing and iteration to achieve satisfactory results.

Furthermore, adapting to new technologies such as TypeScript presented a learning curve as we were unfamiliar with the language. We had to invest time and resources into learning and upskilling to effectively utilize TypeScript in our project.

In addition to these challenges, determining the occurrence of each category of reviews to determine the sentimental analysis of each app was a significant obstacle. Once reviews were clustered, quantifying the frequency of each tag was essential to understanding the distribution of reviews across different aspects of the app.

To address this, we developed a systematic approach to count the occurrences of each category. This involved parsing through the clustered reviews and implementing algorithms to tally the number of reviews associated with each tag. We also accounted for variations in language and expression to ensure accurate categorization and counting.