Embargoed and Restricted Access
Reason: Under embargo until November 2022. After this date a copy can be supplied under Section 51(2) of the Australian Copyright Act 1968 by submitting a document delivery request through your library, or by emailing document.delivery@monash.edu
Improving co-SVD using Contextual Information Imputation
thesis
posted on 2021-11-01, 07:03 authored by JIA MING LOWco-SVD, a recent state-of-the-art algorithm, does not perform well in handling cold-start problems. By integrating the users' pre-use preferences, the quality (F1@5 score) of the top-5 recommendations generated by the co-SVD model can be improved by 5.01% from the baseline under the cold-start user situation. With the long-tail item (cold-start item) situation, the proposed approach can be improved up to 2.39%. The proposed approach is method-agnostic, and other CF-based models can also benefit from it.