The explainability of recommender systems has attracted significant attention in academia and industry. Many efforts have been made for explainable recommendations, yet evaluating the quality of the explanations remains a challenging and unresolved issue. In recent years, leveraging LLMs as evaluators presents a promising avenue in Natural Language Processing tasks (e.g., sentiment classification, information extraction), as they perform strong capabilities in instruction following and common-sense reasoning. However, evaluating recommendation explanatory texts is different from these NLG tasks, as its criteria are related to human perceptions and are usually subjective. In this paper, we investigate whether LLMs can serve as evaluators of recommendation explanations. To answer the question, we utilize real user feedback on explanations given from previous work and additionally collect third-party annotations and LLM evaluations. We design and apply a 3-level meta evaluation strategy to measure the correlation between evaluator labels and the ground truth provided by users. Our experiments reveal that LLMs, such as GPT4, can provide comparable evaluations with appropriate prompts and settings. We also provide further insights into combining human labels with the LLM evaluation process and utilizing ensembles of multiple heterogeneous LLM evaluators to enhance the accuracy and stability of evaluations. Our study verifies that utilizing LLMs as evaluators can be an accurate, reproducible and cost-effective solution for evaluating recommendation explanation texts. Our code is available at https://github.com/Xiaoyu-SZ/LLMasEvaluator.
Simultaneous speech-to-speech translation (Simul-S2ST, a.k.a streaming speech translation) outputs target speech while receiving streaming speech inputs, which is critical for real-time communication. Beyond accomplishing translation between speech, Simul-S2ST requires a policy to control the model to generate corresponding target speech at the opportune moment within speech inputs, thereby posing a double challenge of translation and policy. In this paper, we propose StreamSpeech, a direct Simul-S2ST model that jointly learns translation and simultaneous policy in a unified framework of multi-task learning. Adhering to a multi-task learning approach, StreamSpeech can perform offline and simultaneous speech recognition, speech translation and speech synthesis via an "All-in-One" seamless model. Experiments on CVSS benchmark demonstrate that StreamSpeech achieves state-of-the-art performance in both offline S2ST and Simul-S2ST tasks. Besides, StreamSpeech is able to present high-quality intermediate results (i.e., ASR or translation results) during simultaneous translation process, offering a more comprehensive real-time communication experience.
Multimodal reasoning with large language models (LLMs) often suffers from hallucinations and the presence of deficient or outdated knowledge within LLMs. Some approaches have sought to mitigate these issues by employing textual knowledge graphs, but their singular modality of knowledge limits comprehensive cross-modal understanding. In this paper, we propose the Multimodal Reasoning with Multimodal Knowledge Graph (MR-MKG) method, which leverages multimodal knowledge graphs (MMKGs) to learn rich and semantic knowledge across modalities, significantly enhancing the multimodal reasoning capabilities of LLMs. In particular, a relation graph attention network is utilized for encoding MMKGs and a cross-modal alignment module is designed for optimizing image-text alignment. A MMKG-grounded dataset is constructed to equip LLMs with initial expertise in multimodal reasoning through pretraining. Remarkably, MR-MKG achieves superior performance while training on only a small fraction of parameters, approximately 2.25% of the LLM's parameter size. Experimental results on multimodal question answering and multimodal analogy reasoning tasks demonstrate that our MR-MKG method outperforms previous state-of-the-art models.
Large Language Models (LLMs) have demonstrated an impressive capability known as In-context Learning (ICL), which enables them to acquire knowledge from textual demonstrations without the need for parameter updates. However, many studies have highlighted that the model's performance is sensitive to the choice of demonstrations, presenting a significant challenge for practical applications where we lack prior knowledge of user queries. Consequently, we need to construct an extensive demonstration pool and incorporate external databases to assist the model, leading to considerable time and financial costs. In light of this, some recent research has shifted focus towards zero-shot ICL, aiming to reduce the model's reliance on external information by leveraging their inherent generative capabilities. Despite the effectiveness of these approaches, the content generated by the model may be unreliable, and the generation process is time-consuming. To address these issues, we propose Demonstration Augmentation for In-context Learning (DAIL), which employs the model's previously predicted historical samples as demonstrations for subsequent ones. DAIL brings no additional inference cost and does not rely on the model's generative capabilities. Our experiments reveal that DAIL can significantly improve the model's performance over direct zero-shot inference and can even outperform few-shot ICL without any external information.
Collaborative Filtering (CF) typically suffers from the significant challenge of popularity bias due to the uneven distribution of items in real-world datasets. This bias leads to a significant accuracy gap between popular and unpopular items. It not only hinders accurate user preference understanding but also exacerbates the Matthew effect in recommendation systems. To alleviate popularity bias, existing efforts focus on emphasizing unpopular items or separating the correlation between item representations and their popularity. Despite the effectiveness, existing works still face two persistent challenges: (1) how to extract common supervision signals from popular items to improve the unpopular item representations, and (2) how to alleviate the representation separation caused by popularity bias. In this work, we conduct an empirical analysis of popularity bias and propose Popularity-Aware Alignment and Contrast (PAAC) to address two challenges. Specifically, we use the common supervisory signals modeled in popular item representations and propose a novel popularity-aware supervised alignment module to learn unpopular item representations. Additionally, we suggest re-weighting the contrastive learning loss to mitigate the representation separation from a popularity-centric perspective. Finally, we validate the effectiveness and rationale of PAAC in mitigating popularity bias through extensive experiments on three real-world datasets. Our code is available at https://github.com/miaomiao-cai2/KDD2024-PAAC.
With the applications of recommendation systems rapidly expanding, an increasing number of studies have focused on every aspect of recommender systems with different data inputs, models, and task settings. Therefore, a flexible library is needed to help researchers implement the experimental strategies they require. Existing open libraries for recommendation scenarios have enabled reproducing various recommendation methods and provided standard implementations. However, these libraries often impose certain restrictions on data and seldom support the same model to perform different tasks and input formats, limiting users from customized explorations. To fill the gap, we propose ReChorus2.0, a modular and task-flexible library for recommendation researchers. Based on ReChorus, we upgrade the supported input formats, models, and training&evaluation strategies to help realize more recommendation tasks with more data types. The main contributions of ReChorus2.0 include: (1) Realization of complex and practical tasks, including reranking and CTR prediction tasks; (2) Inclusion of various context-aware and rerank recommenders; (3) Extension of existing and new models to support different tasks with the same models; (4) Support of highly-customized input with impression logs, negative items, or click labels, as well as user, item, and situation contexts. To summarize, ReChorus2.0 serves as a comprehensive and flexible library better aligning with the practical problems in the recommendation scenario and catering to more diverse research needs. The implementation and detailed tutorials of ReChorus2.0 can be found at https://github.com/THUwangcy/ReChorus.
Drama is a form of storytelling inspired by human creativity, proceeding with a predefined storyline, carrying emotions and thoughts. This paper introduces \emph{LLM-based interactive drama}, which endows traditional drama with an unprecedented immersion, where a person is allowed to walk into it and interact with the characters and scenes. We define this new artistic genre by 6 essential elements-plot, character, thought, diction, spectacle and interaction-and study the entire pipeline to forge a backbone \emph{drama LLM} to drive the playing process, which is challenged by limited drama resources, uncontrollable narrative development, and complicated instruction following. We propose \emph{Narrative Chain} to offer finer control over the narrative progression during interaction with players; \emph{Auto-Drama} to synthesize drama scripts given arbitrary stories; \emph{Sparse Instruction Tuning} to allow the model to follow sophisticated instructions. We manually craft 3 scripts, \emph{Detective Conan}, \emph{Harry Potter}, \emph{Romeo and Juliet}, and design a 5-dimension principle to evaluate the drama LLM comprehensively.
The scarcity of non-English data limits the development of non-English large language models (LLMs). Transforming English-centric LLMs to non-English has been identified as an effective and resource-efficient method. Previous works start from base LLMs and perform knowledge distillation (KD) with data generated by stronger LLMs, e.g. GPT-4. Compared to base LLMs, chat LLMs are further optimized for advanced abilities, e.g. multi-turn conversation and human preference alignment, and thus more powerful in both helpfulness and safety. However, transforming a chat LLM involves two critical issues: (1) How can we effectively transfer advanced abilities without their supervised data? (2) How can we prevent the original knowledge from catastrophic forgetting during transformation? We target these issues by introducing a simple framework called TransLLM. For the first issue, TransLLM divides the transfer problem into some common sub-tasks with the translation chain-of-thought, which uses the translation as the bridge between English and non-English step-by-step. We further enhance the performance of sub-tasks with publicly available data. For the second issue, we propose a method comprising two synergistic components: low-rank adaptation for training to maintain the original LLM parameters, and recovery KD, which utilizes data generated by the chat LLM itself to recover the original knowledge from the frozen parameters. In the experiments, we transform the LLaMA-2-chat-7B to the Thai language. Our method, using only single-turn data, outperforms strong baselines and ChatGPT on multi-turn benchmark MT-bench. Furthermore, our method, without safety data, rejects more harmful queries of safety benchmark AdvBench than both ChatGPT and GPT-4.
The rapid advancement of large language models has given rise to a plethora of applications across a myriad of real-world tasks, mainly centered on aligning with human intent. However, the complexities inherent in human intent necessitate a dependence on labor-intensive and time-consuming human evaluation. To alleviate this constraint, we delve into the paradigm of employing open-source large language models as evaluators, aligning with the prevailing trend of utilizing GPT-4. Particularly, we present a step-by-step evaluation framework: \textbf{Fennec}, capable of \textbf{F}ine-grained \textbf{E}valuatio\textbf{N} and correctio\textbf{N} \textbf{E}xtended through bran\textbf{C}hing and bridging. Specifically, the branching operation dissects the evaluation task into various dimensions and granularities, thereby alleviating the challenges associated with evaluation. Concurrently, the bridging operation amalgamates diverse training datasets, augmenting the variety of evaluation tasks. In experimental trials, our 7B model consistently outperforms open-source larger-scale evaluation models across various widely adopted benchmarks in terms of both \textit{Agreement} and \textit{Consistency}, closely approaching the capabilities of GPT-4. We employ the fine-grained correction capabilities induced by the evaluation model to refine multiple model responses, and the results show that the refinement elevates the quality of responses, leading to an improvement of 1-2 points on the MT-Bench. Our code is available at Github\footnote{\url{https://github.com/dropreg/Fennec}}.
As its availability and generality in online services, implicit feedback is more commonly used in recommender systems. However, implicit feedback usually presents noisy samples in real-world recommendation scenarios (such as misclicks or non-preferential behaviors), which will affect precise user preference learning. To overcome the noisy samples problem, a popular solution is based on dropping noisy samples in the model training phase, which follows the observation that noisy samples have higher training losses than clean samples. Despite the effectiveness, we argue that this solution still has limits. (1) High training losses can result from model optimization instability or hard samples, not just noisy samples. (2) Completely dropping of noisy samples will aggravate the data sparsity, which lacks full data exploitation. To tackle the above limitations, we propose a Double Correction Framework for Denoising Recommendation (DCF), which contains two correction components from views of more precise sample dropping and avoiding more sparse data. In the sample dropping correction component, we use the loss value of the samples over time to determine whether it is noise or not, increasing dropping stability. Instead of averaging directly, we use the damping function to reduce the bias effect of outliers. Furthermore, due to the higher variance exhibited by hard samples, we derive a lower bound for the loss through concentration inequality to identify and reuse hard samples. In progressive label correction, we iteratively re-label highly deterministic noisy samples and retrain them to further improve performance. Finally, extensive experimental results on three datasets and four backbones demonstrate the effectiveness and generalization of our proposed framework.