Can we aggregate human intelligence? an approach for human centric aggregation using ordered weighted averaging operators
This work addresses the cold start and storage efficiency issues in recommender systems, particularly for domain-specific applications like education, though it appears incremental as it builds on existing OWA techniques.
The paper tackles the problem of improving recommender systems by prioritizing expert suggestions over general recommendations using a human-centric aggregation approach based on Ordered Weighted Averaging (OWA) operators, applied to book recommendations for university students, with evaluation showing superiority over traditional methods and claims of storage savings and solving the cold start problem.
The primary objective of this paper is to present an approach for recommender systems that can assimilate ranking to the voters or rankers so that recommendation can be made by giving priority to experts suggestion over usual recommendation. To accomplish this, we have incorporated the concept of human-centric aggregation via Ordered Weighted Aggregation (OWA). Here, we are advocating ranked recommendation where rankers are assigned weights according to their place in the ranking. Further, the recommendation process which is presented here for the recommendation of books to university students exploits linguistic data summaries and Ordered Weighted Aggregation (OWA) technique. In the suggested approach, the weights are assigned in a way that it associates higher weights to best ranked university. The approach has been evaluated over eight different parameters. The superiority of the proposed approach is evident from the evaluation results. We claim that proposed scheme saves storage spaces required in traditional recommender systems as well as it does not need users prior preferences and hence produce a solution for cold start problem. This envisaged that the proposed scheme can be very useful in decision making problems, especially for recommender systems. In addition, it emphasizes on how human-centric aggregation can be useful in recommendation researches, and also it gives a new direction about how various human specific tasks can be numerically aggregated.