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

With large language models (LLMs) achieving remarkable breakthroughs innatural language processing (NLP) domains, LLM-enhanced recommender systemshave received much attention and have been actively explored currently. In thispaper, we focus on adapting and empowering a pure large language model forzero-shot and few-shot recommendation tasks. First and foremost, we identifyand formulate the lifelong sequential behavior incomprehension problem for LLMsin recommendation domains, i.e., LLMs fail to extract useful information from atextual context of long user behavior sequence, even if the length of contextis far from reaching the context limitation of LLMs. To address such an issueand improve the recommendation performance of LLMs, we propose a novelframework, namely Retrieval-enhanced Large Language models (ReLLa) forrecommendation tasks in both zero-shot and few-shot settings. For zero-shotrecommendation, we perform semantic user behavior retrieval (SUBR) to improvethe data quality of testing samples, which greatly reduces the difficulty forLLMs to extract the essential knowledge from user behavior sequences. As forfew-shot recommendation, we further design retrieval-enhanced instructiontuning (ReiT) by adopting SUBR as a data augmentation technique for trainingsamples. Specifically, we develop a mixed training dataset consisting of boththe original data samples and their retrieval-enhanced counterparts. We conductextensive experiments on three real-world public datasets to demonstrate thesuperiority of ReLLa compared with existing baseline models, as well as itscapability for lifelong sequential behavior comprehension. To be highlighted,with only less than 10% training samples, few-shot ReLLa can outperformtraditional CTR models that are trained on the entire training set (e.g.,DCNv2, DIN, SIM).

ReLLa: Retrieval-enhanced Large Language Models for Lifelong Sequential Behavior Comprehension in Recommendation