I am looking for some details from different angle (i.e., academic or industrial) about latest Content-based Recommender Systems (RSs) development in any domain, especially E-Marketplace/E-Commerce. It would be nice if you can provide the details with evidence.
It could be one of the following, but not limited to :
Any comments are highly welcomed and appreciated.
Gazillion thanks.
F. M. Bel ́em, C. S. Batista, R. L. T. Santos, J. M. Almeida, and M. A. Gonc ̧alves, “Beyond relevance: Explicitly promoting novelty and diversity in tag recommendation,” ACM Trans. Intell. Syst. Technol., vol. 7, pp. 26:1–26:34, Feb. 2016.
- Adds the element of diversity and novelty within the attributes of tag recommendation. Novelty can be either popularity based or dissimilarity based while diversity can be implicit or explicit.
Dear @Ebenezar,
Thanks for you reply. I found some interesting subjects in your article. Can you please explain how to relate Content-based RSs with the usefulness of tag-based recommendation?
Thanks.
Dear Mohamad,
One of the examples in the paper I've prescribed includes the tagging of Harry Potter movies. The extra tag 'wizard' is added as a result of the novelty present in the film that discriminates itself from the other films in the same genre. This will facilitate in the generation of new tags based on novelty backed up by the popularity that the novel feature welcomes.
Note the main difference between a collaborative filter (CF)-RS and a CB-RS is that the set of features of a standard CF-RS can provide implicit properties that is reflected by the similarities in the user selection of items. But this feature in not present in a CB-RS. Hence the feature set of a CB-RS should be as rich as possible, enough to define user interests that can even change over time.
- Badges
- Science topic
- Similar topics
- Social Science
- Media
- Journalism