As a class project we decided to develop a computational cognitive agent to simulate how people predict the winners of soccer games. Sounds Fancy 🙂
To the best of my knowledge, we first need to observe and extract factors that a human normally takes into account to predict a game between two teams. After extracting all the factors, we need use one of the existing methods (e.g., rule-based model) to model this. Based on my understanding, probably what we need to do is to interview many people. We need to ask people to predict soccer results of a match between two teams (e.g., suppose there is a match between Germany vs. Brazil. Which team do you think will win the match?). We then need to ask them to explain their logic behind their prediction and try to extract factors. For example, people might say Germany will win because it has famous players. Then famous players would be one of our factors. One of the questions that I am not sure about is how do we want to take into account emotional factors and adapt them to our model? For example, what if the person says Germany will win the game because he/she emotionally is attached to this team. How do we want to take such factors into account?
If we could ideally extract the factors that people take into account to predict the games, we would be able to develop our first model. Based on our group discussion we thought that maybe a decision-tree cognitive model could be a good starting point. After collecting the data, we should probably fed our data into a cognitive agent fitted with a decision tree (computational model) that is capable of predicting a winner for given a set of inputs. However, I am not sure how we are going to implement it yet.