Metadata Based Cross-Domain Recommender Framework Using Neighborhood Mapping
- Resource Type
- Conference
- Authors
- Khan, Rabia; Iltaf, Naima; Shafiq, Muhammad Umar; Ur Rehman, Fawad; Obaidullah
- Source
- 2023 International Conference on Sustainable Technology and Engineering (i-COSTE) Sustainable Technology and Engineering (i-COSTE), 2023 International Conference on. :1-8 Dec, 2023
- Subject
- Aerospace
Bioengineering
Components, Circuits, Devices and Systems
Computing and Processing
Engineering Profession
Power, Energy and Industry Applications
Robotics and Control Systems
Transportation
Representation learning
Fuses
Convolution
Soft sensors
Semantics
Coherence
Metadata
Cross-domain recommendation
Collaborative filtering
Graph based representation
Deep learning
Cold-start users
Graph Convolution Networks
- Language
With the overwhelming growth of data there comes a challenge of extracting meaningful and useful information from that data and use it in the most optimized fashion. Modern Recommender Systems (RS) utilize the pattern of user's interest to manage recommendations. However, cold-start users are a daunting challenge in the design of recommender systems since the conventional recommendation services are based on solely one data source. During the recent years, the cross-domain recommendation methods have gained popularity because of the availability of information in multiple domains for cold- start users. The proposed framework, “Metadata based Cross- Domain Recommender Framework using Neighborhood Mapping”(MCDNM) supplements this information by utilizing the data contained in the metadata related to users. This source of information has been mostly ignored by current recommender systems. The rich semantics of metadata are exploited to ex-tract nature of interests of users. The combined advantages of metadata-based and cross-domain approaches are expected to alleviate the issues of cold-start users by transferring user preferences from an auxiliary domain to a target domain. The results demonstrate that the model outperforms the state of the art cross domain recommender systems.