A Bi-Step Grounding Paradigm for Large Language Models in Recommendation Systems
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2023
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866917557122891776 |
|---|---|
| author | Bao, Keqin Zhang, Jizhi Wang, Wenjie Zhang, Yang Yang, Zhengyi Luo, Yancheng Chen, Chong Feng, Fuli Tian, Qi |
| author_facet | Bao, Keqin Zhang, Jizhi Wang, Wenjie Zhang, Yang Yang, Zhengyi Luo, Yancheng Chen, Chong Feng, Fuli Tian, Qi |
| contents | As the focus on Large Language Models (LLMs) in the field of recommendation intensifies, the optimization of LLMs for recommendation purposes (referred to as LLM4Rec) assumes a crucial role in augmenting their effectiveness in providing recommendations. However, existing approaches for LLM4Rec often assess performance using restricted sets of candidates, which may not accurately reflect the models' overall ranking capabilities. In this paper, our objective is to investigate the comprehensive ranking capacity of LLMs and propose a two-step grounding framework known as BIGRec (Bi-step Grounding Paradigm for Recommendation). It initially grounds LLMs to the recommendation space by fine-tuning them to generate meaningful tokens for items and subsequently identifies appropriate actual items that correspond to the generated tokens. By conducting extensive experiments on two datasets, we substantiate the superior performance, capacity for handling few-shot scenarios, and versatility across multiple domains exhibited by BIGRec. Furthermore, we observe that the marginal benefits derived from increasing the quantity of training samples are modest for BIGRec, implying that LLMs possess the limited capability to assimilate statistical information, such as popularity and collaborative filtering, due to their robust semantic priors. These findings also underline the efficacy of integrating diverse statistical information into the LLM4Rec framework, thereby pointing towards a potential avenue for future research. Our code and data are available at https://github.com/SAI990323/Grounding4Rec. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2308_08434 |
| institution | arXiv |
| publishDate | 2023 |
| record_format | arxiv |
| spellingShingle | A Bi-Step Grounding Paradigm for Large Language Models in Recommendation Systems Bao, Keqin Zhang, Jizhi Wang, Wenjie Zhang, Yang Yang, Zhengyi Luo, Yancheng Chen, Chong Feng, Fuli Tian, Qi Information Retrieval As the focus on Large Language Models (LLMs) in the field of recommendation intensifies, the optimization of LLMs for recommendation purposes (referred to as LLM4Rec) assumes a crucial role in augmenting their effectiveness in providing recommendations. However, existing approaches for LLM4Rec often assess performance using restricted sets of candidates, which may not accurately reflect the models' overall ranking capabilities. In this paper, our objective is to investigate the comprehensive ranking capacity of LLMs and propose a two-step grounding framework known as BIGRec (Bi-step Grounding Paradigm for Recommendation). It initially grounds LLMs to the recommendation space by fine-tuning them to generate meaningful tokens for items and subsequently identifies appropriate actual items that correspond to the generated tokens. By conducting extensive experiments on two datasets, we substantiate the superior performance, capacity for handling few-shot scenarios, and versatility across multiple domains exhibited by BIGRec. Furthermore, we observe that the marginal benefits derived from increasing the quantity of training samples are modest for BIGRec, implying that LLMs possess the limited capability to assimilate statistical information, such as popularity and collaborative filtering, due to their robust semantic priors. These findings also underline the efficacy of integrating diverse statistical information into the LLM4Rec framework, thereby pointing towards a potential avenue for future research. Our code and data are available at https://github.com/SAI990323/Grounding4Rec. |
| title | A Bi-Step Grounding Paradigm for Large Language Models in Recommendation Systems |
| topic | Information Retrieval |
| url | https://arxiv.org/abs/2308.08434 |