arXiv subjects

Pu-Jen Cheng

Publications and source records attributed to Pu-Jen Cheng.

18 recordsLinked to original sources

arXiv2026

Are We Really Making Progress in Group Recommendation? Unmasking the Tie-Breaking Illusion

Recent group recommendation methods have reported strong improvements on standard benchmarks, but it remains unclear whether these gains always reflect genuine advances in modeling group preferences. In this paper, we show that several recent methods are affected by a systematic evaluation bias caused by the interaction between training-time score compression and evaluation-time deterministic tie-breaking. Specifically, an additional sigmoid transformation before the BPR objective can greatly increase tied top scores, making top-K metrics such as HR@K and NDCG@K highly sensitive to how ties are resolved. We revisit recent representative methods and their baselines on CAMRa2011 and Mafengwo under both group and user recommendation settings, and evaluate them with a tie-aware protocol that computes the exact expectation of HR@K and NDCG@K under uniform random tie-breaking. Our results show that many previously reported improvements shrink substantially under tie-aware evaluation, and the relative ranking of methods can change markedly. We further show that the additional sigmoid may act as implicit margin smoothing during optimization, and that temperature-scaled BPR can retain much of this benefit without inducing severe tie inflation. Overall, our findings highlight the importance of tie-aware evaluation for establishing reliable progress in group recommendation. The code is available at https://github.com/songduoma/TieAwareGroupRec.

cs.IR↗

arXiv2026

Structure-Preserving Projection for Mitigating Modality Bias in LLM-Based Sequential Recommendation

Recent LLM-based recommenders integrate textual and collaborative signals by projecting collaborative embeddings into the embedding space of the LLM. However, this projection can introduce modality bias that distorts the underlying collaborative structure and limits the usefulness of projected embeddings. To address this issue, we propose a novel structure-preserving projection approach that maintains the relational geometry of collaborative embeddings through dedicated structure-preserving losses. Comprehensive experiments demonstrate that our approach consistently improves recommendation performance, providing a more reliable path for LLM-based recommendation.

cs.IR↗

arXiv2026

ICICLE: Expanding Retrieval with In-Context Documents

Generative retrieval (GR) maps queries directly to document identifiers (docids) using parametric knowledge, However, this design makes corpus expansion costly: adding new documents requires updating model parameters to encode new document-docid associations incurs repeated training and catastrophic forgetting of previously indexed documents. In this work, we revisit incremental GR as an in-context retrieval problem, where newly added documents are supplied as inference-time document-docid evidence. We propose ICICLE, an in-context indexing framework that performs source-aware docid generation over both parametric memory and context-provided document-docid pairs. ICICLE combines a `[COPY]`-based routing mechanism, preference-based calibration, and large context adaptation to distinguish context-grounded retrieval from parametric retrieval. Experiments on MS MARCO and NQ320K show that ICICLE improves retrieval of newly introduced documents while preserving seen-document retention without corpus-specific retraining. Our analysis further shows that high-shot degradation is mainly caused by routing failure, highlighting source-selection calibration as a key bottleneck for scaling in-context generative retrieval.

cs.IR↗

arXiv2026

Beyond Uniform Token Training: A Multi-Target Framework for Learning Token-Weighted Objectives in Generative Recommenders

Recent generative recommendation models recast next-item prediction as the generation of a semantic identifier sequence. While this formulation enables autoregressive models to produce item IDs directly, the commonly used token-level likelihood objective does not distinguish between tokens that play different roles in item identification. This limitation is especially pronounced for semantic-ID representations, where prefix tokens often determine coarse item groups and later tokens provide finer-grained disambiguation. To better align training with the structure of semantic IDs, we study token-level learning signals from two complementary perspectives. First, we introduce a prefix-aware weighting scheme, Front-Greater Weighting, which emphasizes tokens according to their contribution to reducing semantic ambiguity among candidate items. Second, frequency weighting increases the learning emphasis on infrequent tokens, addressing the long-tailed distributions and popularity bias commonly observed in recommendation data. We further introduce a multi-target optimization framework with curriculum learning, which integrates the two token-weighted objectives with the standard likelihood and enables stable optimization with adaptive emphasis across training stages. Experiments on multiple benchmark datasets demonstrate that the proposed approach consistently improves generative recommendation performance over strong baselines and prior token-weighting methods. Additional analyses show that the method is robust across different semantic-ID constructions and backbone scales, and that it improves recommendation quality for both popular and long-tail items. Code is available at github repository.

cs.IR↗

arXiv2025

Doc2Query++: Topic-Coverage based Document Expansion and its Application to Dense Retrieval via Dual-Index Fusion

Document expansion (DE) via query generation tackles vocabulary mismatch in sparse retrieval, yet faces limitations: uncontrolled generation producing hallucinated or redundant queries with low diversity; poor generalization from in-domain training (e.g., MS MARCO) to out-of-domain data like BEIR; and noise from concatenation harming dense retrieval. While Large Language Models (LLMs) enable cross-domain query generation, basic prompting lacks control, and taxonomy-based methods rely on domain-specific structures, limiting applicability. To address these challenges, we introduce Doc2Query++, a DE framework that structures query generation by first inferring a document's latent topics via unsupervised topic modeling for cross-domain applicability, then using hybrid keyword selection to create a diverse and relevant keyword set per document. This guides LLM not only to leverage keywords, which ensure comprehensive topic representation, but also to reduce redundancy through diverse, relevant terms. To prevent noise from query appending in dense retrieval, we propose Dual-Index Fusion strategy that isolates text and query signals, boosting performance in dense settings. Extensive experiments show Doc2Query++ significantly outperforms state-of-the-art baselines, achieving substantial gains in MAP, nDCG@10 and Recall@100 across diverse datasets on both sparse and dense retrieval.

cs.IR↗

arXiv2025

Beyond Independent Passages: Adaptive Passage Combination Retrieval for Retrieval Augmented Open-Domain Question Answering

Retrieval-augmented generation (RAG) enhances large language models (LLMs) by incorporating external documents at inference time, enabling up-to-date knowledge access without costly retraining. However, conventional RAG methods retrieve passages independently, often leading to redundant, noisy, or insufficiently diverse context-particularly problematic - particularly problematic in noisy corpora and for multi-hop questions. To address this, we propose Adaptive Passage Combination Retrieval (AdaPCR), a novel framework for open-domain question answering with black-box LMs. AdaPCR explicitly models dependencies between passages by considering passage combinations as units for retrieval and reranking. It consists of a context-aware query reformulation using concatenated passages, and a reranking step trained with a predictive objective aligned with downstream answer likelihood. Crucially, AdaPCR adaptively selects the number of retrieved passages without additional stopping modules. Experiments across several QA benchmarks show that AdaPCR outperforms baselines, particularly in multi-hop reasoning, demonstrating the effectiveness of modeling inter-passage dependencies for improved retrieval.

cs.CL↗

arXiv2025

Augment or Not? A Comparative Study of Pure and Augmented Large Language Model Recommenders

Large language models (LLMs) have introduced new paradigms for recommender systems by enabling richer semantic understanding and incorporating implicit world knowledge. In this study, we propose a systematic taxonomy that classifies existing approaches into two categories: (1) Pure LLM Recommenders, which rely solely on LLMs, and (2) Augmented LLM Recommenders, which integrate additional non-LLM techniques to enhance performance. This taxonomy provides a novel lens through which to examine the evolving landscape of LLM-based recommendation. To support fair comparison, we introduce a unified evaluation platform that benchmarks representative models under consistent experimental settings, highlighting key design choices that impact effectiveness. We conclude by discussing open challenges and outlining promising directions for future research. This work offers both a comprehensive overview and practical guidance for advancing next-generation LLM-powered recommender.

cs.IR↗

arXiv2024

Not All LLM-Generated Data Are Equal: Rethinking Data Weighting in Text Classification

Synthetic data augmentation via large language models (LLMs) allows researchers to leverage additional training data, thus enhancing the performance of downstream tasks, especially when real-world data is scarce. However, the generated data can deviate from the real-world data, and this misalignment can bring deficient outcomes while applying the trained model to applications. Therefore, we proposed efficient weighted-loss approaches to align synthetic data with real-world distribution by emphasizing high-quality and diversified data generated by LLMs with using merely a little real-world data. We empirically assessed the effectiveness of our method on multiple text classification tasks, and the results showed leveraging our approaches on a BERT-level model robustly outperformed standard cross-entropy and other data weighting approaches, providing potential solutions to effectively leveraging synthetic data from any suitable data generator for model training.

cs.LG↗

arXiv2024

Legal Documents Drafting with Fine-Tuned Pre-Trained Large Language Model

With the development of large-scale Language Models (LLM), fine-tuning pre-trained LLM has become a mainstream paradigm for solving downstream tasks of natural language processing. However, training a language model in the legal field requires a large number of legal documents so that the language model can learn legal terminology and the particularity of the format of legal documents. The typical NLP approaches usually rely on many manually annotated data sets for training. However, in the legal field application, it is difficult to obtain a large number of manually annotated data sets, which restricts the typical method applied to the task of drafting legal documents. The experimental results of this paper show that not only can we leverage a large number of annotation-free legal documents without Chinese word segmentation to fine-tune a large-scale language model, but more importantly, it can fine-tune a pre-trained LLM on the local computer to achieve the generating legal document drafts task, and at the same time achieve the protection of information privacy and to improve information security issues.

cs.CL↗

arXiv2024

Improving One-class Recommendation with Multi-tasking on Various Preference Intensities

In the one-class recommendation problem, it's required to make recommendations basing on users' implicit feedback, which is inferred from their action and inaction. Existing works obtain representations of users and items by encoding positive and negative interactions observed from training data. However, these efforts assume that all positive signals from implicit feedback reflect a fixed preference intensity, which is not realistic. Consequently, representations learned with these methods usually fail to capture informative entity features that reflect various preference intensities. In this paper, we propose a multi-tasking framework taking various preference intensities of each signal from implicit feedback into consideration. Representations of entities are required to satisfy the objective of each subtask simultaneously, making them more robust and generalizable. Furthermore, we incorporate attentive graph convolutional layers to explore high-order relationships in the user-item bipartite graph and dynamically capture the latent tendencies of users toward the items they interact with. Experimental results show that our method performs better than state-of-the-art methods by a large margin on three large-scale real-world benchmark datasets.

cs.IR↗

arXiv2024

Learning Unsupervised Semantic Document Representation for Fine-grained Aspect-based Sentiment Analysis

Document representation is the core of many NLP tasks on machine understanding. A general representation learned in an unsupervised manner reserves generality and can be used for various applications. In practice, sentiment analysis (SA) has been a challenging task that is regarded to be deeply semantic-related and is often used to assess general representations. Existing methods on unsupervised document representation learning can be separated into two families: sequential ones, which explicitly take the ordering of words into consideration, and non-sequential ones, which do not explicitly do so. However, both of them suffer from their own weaknesses. In this paper, we propose a model that overcomes difficulties encountered by both families of methods. Experiments show that our model outperforms state-of-the-art methods on popular SA datasets and a fine-grained aspect-based SA by a large margin.

cs.LG↗

arXiv2023

Program Machine Policy: Addressing Long-Horizon Tasks by Integrating Program Synthesis and State Machines

Deep reinforcement learning (deep RL) excels in various domains but lacks generalizability and interpretability. On the other hand, programmatic RL methods (Trivedi et al., 2021; Liu et al., 2023) reformulate RL tasks as synthesizing interpretable programs that can be executed in the environments. Despite encouraging results, these methods are limited to short-horizon tasks. On the other hand, representing RL policies using state machines (Inala et al., 2020) can inductively generalize to long-horizon tasks; however, it struggles to scale up to acquire diverse and complex behaviors. This work proposes the Program Machine Policy (POMP), which bridges the advantages of programmatic RL and state machine policies, allowing for the representation of complex behaviors and the address of long-term tasks. Specifically, we introduce a method that can retrieve a set of effective, diverse, and compatible programs. Then, we use these programs as modes of a state machine and learn a transition function to transition among mode programs, allowing for capturing repetitive behaviors. Our proposed framework outperforms programmatic RL and deep RL baselines on various tasks and demonstrates the ability to inductively generalize to even longer horizons without any fine-tuning. Ablation studies justify the effectiveness of our proposed search algorithm for retrieving a set of programs as modes.

cs.LG↗

arXiv2023

printf: Preference Modeling Based on User Reviews with Item Images and Textual Information via Graph Learning

Nowadays, modern recommender systems usually leverage textual and visual contents as auxiliary information to predict user preference. For textual information, review texts are one of the most popular contents to model user behaviors. Nevertheless, reviews usually lose their shine when it comes to top-N recommender systems because those that solely utilize textual reviews as features struggle to adequately capture the interaction relationships between users and items. For visual one, it is usually modeled with naive convolutional networks and also hard to capture high-order relationships between users and items. Moreover, previous works did not collaboratively use both texts and images in a proper way. In this paper, we propose printf, preference modeling based on user reviews with item images and textual information via graph learning, to address the above challenges. Specifically, the dimension-based attention mechanism directs relations between user reviews and interacted items, allowing each dimension to contribute different importance weights to derive user representations. Extensive experiments are conducted on three publicly available datasets. The experimental results demonstrate that our proposed printf consistently outperforms baseline methods with the relative improvements for NDCG@5 of 26.80%, 48.65%, and 25.74% on Amazon-Grocery, Amazon-Tools, and Amazon-Electronics datasets, respectively. The in-depth analysis also indicates the dimensions of review representations definitely have different topics and aspects, assisting the validity of our model design.

cs.IR↗

arXiv2023

Attentive Graph-based Text-aware Preference Modeling for Top-N Recommendation

Textual data are commonly used as auxiliary information for modeling user preference nowadays. While many prior works utilize user reviews for rating prediction, few focus on top-N recommendation, and even few try to incorporate item textual contents such as title and description. Though delivering promising performance for rating prediction, we empirically find that many review-based models cannot perform comparably well on top-N recommendation. Also, user reviews are not available in some recommendation scenarios, while item textual contents are more prevalent. On the other hand, recent graph convolutional network (GCN) based models demonstrate state-of-the-art performance for top-N recommendation. Thus, in this work, we aim to further improve top-N recommendation by effectively modeling both item textual content and high-order connectivity in user-item graph. We propose a new model named Attentive Graph-based Text-aware Recommendation Model (AGTM). Extensive experiments are provided to justify the rationality and effectiveness of our model design.

cs.IR↗

arXiv2023

Hierarchical Programmatic Reinforcement Learning via Learning to Compose Programs

Aiming to produce reinforcement learning (RL) policies that are human-interpretable and can generalize better to novel scenarios, Trivedi et al. (2021) present a method (LEAPS) that first learns a program embedding space to continuously parameterize diverse programs from a pre-generated program dataset, and then searches for a task-solving program in the learned program embedding space when given a task. Despite the encouraging results, the program policies that LEAPS can produce are limited by the distribution of the program dataset. Furthermore, during searching, LEAPS evaluates each candidate program solely based on its return, failing to precisely reward correct parts of programs and penalize incorrect parts. To address these issues, we propose to learn a meta-policy that composes a series of programs sampled from the learned program embedding space. By learning to compose programs, our proposed hierarchical programmatic reinforcement learning (HPRL) framework can produce program policies that describe out-of-distributionally complex behaviors and directly assign credits to programs that induce desired behaviors. The experimental results in the Karel domain show that our proposed framework outperforms baselines. The ablation studies confirm the limitations of LEAPS and justify our design choices.

cs.LG↗

arXiv2022

Mr. Right: Multimodal Retrieval on Representation of ImaGe witH Text

Multimodal learning is a recent challenge that extends unimodal learning by generalizing its domain to diverse modalities, such as texts, images, or speech. This extension requires models to process and relate information from multiple modalities. In Information Retrieval, traditional retrieval tasks focus on the similarity between unimodal documents and queries, while image-text retrieval hypothesizes that most texts contain the scene context from images. This separation has ignored that real-world queries may involve text content, image captions, or both. To address this, we introduce Multimodal Retrieval on Representation of ImaGe witH Text (Mr. Right), a novel and comprehensive dataset for multimodal retrieval. We utilize the Wikipedia dataset with rich text-image examples and generate three types of text-based queries with different modality information: text-related, image-related, and mixed. To validate the effectiveness of our dataset, we provide a multimodal training paradigm and evaluate previous text retrieval and image retrieval frameworks. The results show that proposed multimodal retrieval can improve retrieval performance, but creating a well-unified document representation with texts and images is still a challenge. We hope Mr. Right allows us to broaden current retrieval systems better and contributes to accelerating the advancement of multimodal learning in the Information Retrieval.

cs.IR↗

arXiv2022

An Evaluation Dataset for Legal Word Embedding: A Case Study On Chinese Codex

Word embedding is a modern distributed word representations approach widely used in many natural language processing tasks. Converting the vocabulary in a legal document into a word embedding model facilitates subjecting legal documents to machine learning, deep learning, and other algorithms and subsequently performing the downstream tasks of natural language processing vis-\`a-vis, for instance, document classification, contract review, and machine translation. The most common and practical approach of accuracy evaluation with the word embedding model uses a benchmark set with linguistic rules or the relationship between words to perform analogy reasoning via algebraic calculation. This paper proposes establishing a 1,134 Legal Analogical Reasoning Questions Set (LARQS) from the 2,388 Chinese Codex corpus using five kinds of legal relations, which are then used to evaluate the accuracy of the Chinese word embedding model. Moreover, we discovered that legal relations might be ubiquitous in the word embedding model.

cs.CL↗

arXiv2021

End-to-End Video Question-Answer Generation with Generator-Pretester Network

We study a novel task, Video Question-Answer Generation (VQAG), for challenging Video Question Answering (Video QA) task in multimedia. Due to expensive data annotation costs, many widely used, large-scale Video QA datasets such as Video-QA, MSVD-QA and MSRVTT-QA are automatically annotated using Caption Question Generation (CapQG) which inputs captions instead of the video itself. As captions neither fully represent a video, nor are they always practically available, it is crucial to generate question-answer pairs based on a video via Video Question-Answer Generation (VQAG). Existing video-to-text (V2T) approaches, despite taking a video as the input, only generate a question alone. In this work, we propose a novel model Generator-Pretester Network that focuses on two components: (1) The Joint Question-Answer Generator (JQAG) which generates a question with its corresponding answer to allow Video Question "Answering" training. (2) The Pretester (PT) verifies a generated question by trying to answer it and checks the pretested answer with both the model's proposed answer and the ground truth answer. We evaluate our system with the only two available large-scale human-annotated Video QA datasets and achieves state-of-the-art question generation performances. Furthermore, using our generated QA pairs only on the Video QA task, we can surpass some supervised baselines. We apply our generated questions to Video QA applications and surpasses some supervised baselines using generated questions only. As a pre-training strategy, we outperform both CapQG and transfer learning approaches when employing semi-supervised (20%) or fully supervised learning with annotated data. These experimental results suggest the novel perspectives for Video QA training.

cs.MM↗