אפקה - פרויקטי גמר 2020

Gamer Time

Technological Solution

The solution includes a mobile application implemented with Flutter, a

server application implemented with Node.js, an artificial neural

network model implemented with TensorFlow, the use of Firebase

services to allow secure and user-friendly authentication as well as a

realtime database. In addition, the server is interfacing with an

external API in order to fetch information regarding existing games.

Gabriel Kol

Advisor: Perry Shalom

Software Engineering

Recommendations Algorithm

The system recommends games to the user based on their personality

type and actions of other players in their circle of friends. The user’s

personality is

analyzed from texts written in their

reviews. The reviews are being

encoded into 512-dimensional

embeddings vectors which are

used as inputs to our pre-trained

text classification neural network model.

Application Screens

A mobile application that provides a personalized video game

recommendations system, game rating and review

capabilities, social networking service, game search engine,

tracking list management, and more. Through machine

learning and principles of psychology, the system is able to

compute the user's personality type and recommend popular

games in matching genres.

Client

(Flutter Mobile Application)

Server

(Node.js Application)

Realtime Database

(Cloud Firestore)

Games API

(RAWG API)

Neural Network

(TensorFlow)