A satisfaction balance strategy for tourism group recommendation. This strategy combined mean value strategy and minimum pain strategy in the process of fusion to improve the recommendation satisfaction of group members.
A hybrid integration strategy, differences in both sides of the threshold, respectively, adopt the strategy of the most respected person and mean complete preference fusion strategy. However, the selection of threshold value often depends on the situation.
A preference prediction algorithm based on the theory of unknown preferences of users in the same group would be affected by other members in the group. However, the algorithm had a high time complexity.
A method for combining trust in social networks to modify group members’ preferences, but it is usually difficult to obtain trust, so this method is not easy to implement.
The method added the relationship between group members to the joint probability matrix decomposition; it improves the accuracy of cluster recommendation.
The method added weight to candidate projects by calculating the number and consistency of project scores of group members and then made group recommendation by integrating project weights.
A matrix decomposition model. It is combined with the timing function to improve the search completeness and accuracy of the group recommendation system.
A nonnegative matrix decomposition model. It is widely used in computer vision and data mining due to its simplicity of implementation, interpretability of decomposition form, and decomposition result.
A group preference model by weighted fusion of member scores based on the contribution degree of members. However, it lacks consideration of the relationship between the user’s knowledge background and the inherent properties of the item in the preference fusion process.
