Document Type : Primary Research paper
Research Scholar, Department of CSE, PSG College of Technology, Coimbatore, Tamil Nadu, India 641004.
Professor, Department of CSE, PSG College of Technology, Coimbatore, Tamil Nadu, India 641004.
Mobile applications in the app store have increased rapidly over the years and with the increasing popularity, the mobile app developers resist getting visibility for their product. An important factor that influences the visibility of an app is how it gets categorized in the app market. A study was made to identify misclassified apps and categorize them to help app users. To uncover the misclassified mobile apps in the app, store a new approach to categorize the related apps together based on their description and API calls has been proposed. A dataset containing 25,000+ mobile apps mined from the Google Play Store were used. The initial step involves grouping the applications into various categories based on the technical description of the mobile applications. Pre-processing of descriptions was done using natural language processing techniques and feature extraction using Latent Dirichlet Allocation (LDA). The work can be split into two halves. Work I focus on clustering which was again carried out in two methods by varying the parameters, Model A was using the features extracted from app descriptions as a parameter, and Model B was using the features extracted from descriptions and API calls as parameters. K-Mean clustering was used as a clustering technique due to its hard-clustering nature. Both the clustered outputs were evaluated and an efficient one was identified. Work II focuses on classifying the app based on app description. Popular machine learning and deep learning models were used for classification and a comparative study was made.