Development of a crawler to collect online game playing traces

Abstract

As multiplayer online games become a more popular way in public entertainment, data generated in online gaming traces also become a more interesting data source to analysis on. To retrieve those data from online gaming, a web crawler is a common and practical way for data analysis. However, existing web crawlers available online are mostly focusing on document searching in text or metadata, which is not fully applicable in gaming data retrieval. The primary objective of this project is to develop a web crawler to collect real game playing traces. The online game focused in this application is League of Legends, which is known as one of the most famous multi-player online games in the world. To retrieve the data related to League of Legends, Riot Games provides official APIs for developers to play on. The crawler is implemented to view and manipulate on the data retrieve via Riot Games APIs. On the graphical user interface implemented, the user is able to search and view a player’s information, update a player’s information and view overall champion statistics with optional filters. A player’s information includes player profile, league position information, champion masteries and recent matches information. Updated player information is stored in database including the player information and recent matches information. Overall champion statistics covers the win rate, KDA rate, number of games played and average gold for each champion that can be played in League of Legends. In addition, a simple analysis of players’ characteristics is performed by the application as well. Every single player is analyzed by his/her abilities in Carry, Teamwork, Support, Farm, Survive and All-rounder. Based on the six characteristics, K-means clustering is also performed to group the player in 3 different clusters for understanding the role of player.