RecInDial: A Unified Framework for Conversational Recommendation with Pretrained Language Models

Lingzhi Wang,Huang Hu,Lei Sha,Can Xu,Kam-Fai Wong,Daxin Jiang

DOI: https://doi.org/10.48550/arXiv.2110.07477

2022-10-07

Abstract:Conversational Recommender System (CRS), which aims to recommend high-quality items to users through interactive conversations, has gained great research interest recently. A CRS is usually composed of a recommendation module and a generation module. In the previous work, these two modules are loosely connected in the model training and are shallowly integrated during inference, where a simple switching or copy mechanism is adopted to incorporate recommended items into generated responses. Moreover, the current end-to-end neural models trained on small crowd-sourcing datasets (e.g., 10K dialogs in the ReDial dataset) tend to overfit and have poor chit-chat ability. In this work, we propose a novel unified framework that integrates recommendation into the dialog (RecInDial) generation by introducing a vocabulary pointer. To tackle the low-resource issue in CRS, we finetune the large-scale pretrained language models to generate fluent and diverse responses, and introduce a knowledge-aware bias learned from an entity-oriented knowledge graph to enhance the recommendation performance. Furthermore, we propose to evaluate the CRS models in an end-to-end manner, which can reflect the overall performance of the entire system rather than the performance of individual modules, compared to the separate evaluations of the two modules used in previous work. Experiments on the benchmark dataset ReDial show our RecInDial model significantly surpasses the state-of-the-art methods. More extensive analyses show the effectiveness of our model.

Computation and Language

What problem does this paper attempt to address?

The problems that this paper attempts to solve mainly focus on several key challenges encountered in the Conversational Recommender System (CRS):

1. **Shallow integration between modules**: Existing CRSs are usually composed of a recommendation module and a generation module. These two modules are loosely connected during model training, and during the inference process, the recommended items are only incorporated into the generated responses through simple switching or copying mechanisms. This strategy cannot accurately and appropriately integrate the recommended items into the generated responses.

2. **Problem of resource scarcity**: Current end - to - end neural models are prone to over - fitting when trained on small crowdsourced datasets (for example, there are only about 10,000 conversations in the ReDial dataset), and perform poorly in terms of small - talk ability.

3. **Incomplete evaluation methods**: Previous works usually evaluate the performance of the two modules of dialogue generation and item recommendation separately, without evaluating the final recommendation effect of the system as a whole. This leads to a significant gap between the performance of the recommendation module and the final recommendation effect.

To solve the above problems, the paper proposes a unified framework based on pre - trained language models - RecInDial. This framework incorporates recommended items into the dialogue generation process by introducing a vocabulary pointer, uses large - scale pre - trained language models to generate fluent and diverse responses, and enhances the recommendation performance by learning knowledge - aware biases from entity - oriented knowledge graphs. In addition, the paper also proposes to evaluate CRS models in an end - to - end manner to reflect the comprehensive performance of the entire system, rather than just evaluating the performance of each module.

Specifically, the main contributions of RecInDial include:
- Proposing a framework RecInDial based on pre - trained language models, which fine - tunes large - scale pre - trained language models and Relational Graph Convolutional Networks (RGCN) to address the resource scarcity problem in current CRSs.
- Effectively unifying the two components of item recommendation and response generation into a pre - trained language model - based framework by introducing an additional item vocabulary and a vocabulary pointer.
- Extensive experiments show that RecInDial significantly outperforms the existing state - of - the - art methods in both dialogue generation and recommendation evaluations.

These improvements not only improve the accuracy of recommendations but also enhance the naturalness and diversity of conversations, thereby enhancing the user experience.