VertiKin is an e-commerce platform that allows the user to search through an online product inventory. It is also able to automatically detect what users might be interested in buying.

VertiKin Mobile app learns from user inputs on the mobile device (we do not read passwords and private information, so the user can be assured of his or her security). User data is then sent to the VertiKin server and analyzed with natural language processing (NLP). NLP identifies key information, especially frequency, to predict potential product interests. If VertiKin identifies an interest, the server sends a GCM push notification to the user.

If VertiKin incorrectly gauged user interest in a product, the user can offer feedback by pressing No on an in-app form. This feedback is then used to improve further predictions. Users start with a DEFAULT_THRESHOLD the THRESHOLD_DELTA adjusts over time in response to feedback.

• According to a 2010 Nielsen Report, users often discuss product purchases online. We used this to better predict future purchases.
• Cognitive fluency is the human tendency to prefer things that are familiar and easy to understand. According to this article published on boston.com, users prefer easy-to-grasp products. Using this information and knowledge of peer group dynamics, VertiKin can predict that consumers are likely to discuss purchases with family and friends.

• Custom Keyboard by Mehdi Mirzaei
• nltk
• Google Cloud Messaging
• Walmart Open API