English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Zendy Plus

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Zendy Tools

Zendy Open

In this perspective paper we study the effect of non independent andidentically distributed (non-IID) data on federated online learning to rank(FOLTR) and chart directions for future work in this new and largely unexploredresearch area of Information Retrieval. In the FOLTR process, clientsparticipate in a federation to jointly create an effective ranker from theimplicit click signal originating in each client, without the need to sharedata (documents, queries, clicks). A well-known factor that affects theperformance of federated learning systems, and that poses serious challenges tothese approaches, is that there may be some type of bias in the way data isdistributed across clients. While FOLTR systems are on their own rights a typeof federated learning system, the presence and effect of non-IID data in FOLTRhas not been studied. To this aim, we first enumerate possible datadistribution settings that may showcase data bias across clients and thus giverise to the non-IID problem. Then, we study the impact of each setting on theperformance of the current state-of-the-art FOLTR approach, the FederatedPairwise Differentiable Gradient Descent (FPDGD), and we highlight which datadistributions may pose a problem for FOLTR methods. We also explore how commonapproaches proposed in the federated learning literature address non-IID issuesin FOLTR. This allows us to unveil new research gaps that, we argue, futureresearch in FOLTR should consider. This is an important contribution to thecurrent state of FOLTR field because, for FOLTR systems to be deployed, thefactors affecting their performance, including the impact of non-IID data, needto be thoroughly understood.

Is Non-IID Data a Threat in Federated Online Learning to Rank?