Defense of the Ancients (DOTA 2)-Draft Recommendation System

In the last decade, Dota 2 has gradually developed by attracting a lot of players globally and has recently hosted its 10^th international tournament for professional players. Though the data from these matches are easily available on various sites, they cannot be used to implement data mining and deep learning techniques directly. After sufficient data engineering, one can use the data and unleash the benefits that can be created for both, the existing and the new players, which is the objective of our project. The Dota2 game is played in 2 stages: (i) The Drafting Phase (ii) The Playing Phase. The drafting phase is analogous to the team selection phase of the game which has a huge impact on the outcome of the game i.e. a stronger team draft has a better chance of winning. Both teams have to choose and ban heroes/characters in a chronological order and the aim is to build a team that synergizes well with its own team and counters the opponent team. However, it is observed that most players are often unable to pick the best team lineup to compete. The aim is to solve this problem and in this research paper, a draft recommendation system based on the NLP (Natural Language Processing) technique of finding the next word is proposed and thus word2vec algorithm, more specifically the CBOW (Continuous Bag Of Words) algorithm in tandem with LSTM (Long Short-Term Memory) model is used to give out the best suggestions.