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/

Global: Zendy Plus annual

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

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Large language models (LLMs) have demonstrated remarkable capabilities andhave been extensively deployed across various domains, including recommendersystems. Numerous studies have employed specialized \textit{prompts} to harnessthe in-context learning capabilities intrinsic to LLMs. For example, LLMs areprompted to act as zero-shot rankers for listwise ranking, evaluating candidateitems generated by a retrieval model for recommendation. Recent researchfurther uses instruction tuning techniques to align LLM with human preferencefor more promising recommendations. Despite its potential, current researchoverlooks the integration of multiple ranking tasks to enhance modelperformance. Moreover, the signal from the conventional recommendation model isnot integrated into the LLM, limiting the current system performance.  In this paper, we introduce RecRanker, tailored for instruction tuning LLM toserve as the \textbf{Ranker} for top-\textit{k} \textbf{Rec}ommendations.Specifically, we introduce importance-aware sampling, clustering-basedsampling, and penalty for repetitive sampling for sampling high-quality,representative, and diverse training data. To enhance the prompt, we introduceposition shifting strategy to mitigate position bias and augment the promptwith auxiliary information from conventional recommendation models, therebyenriching the contextual understanding of the LLM. Subsequently, we utilize thesampled data to assemble an instruction-tuning dataset with the augmentedprompt comprising three distinct ranking tasks: pointwise, pairwise, andlistwise rankings. We further propose a hybrid ranking method to enhance themodel performance by ensembling these ranking tasks. Our empirical evaluationsdemonstrate the effectiveness of our proposed RecRanker in both direct andsequential recommendation scenarios.

RecRanker: Instruction Tuning Large Language Model as Ranker for Top-k Recommendation