The Hypencoder, proposed by Killingback et al., is a retrieval framework that replaces the fixed inner-product scoring function used in standard bi-encoders with a query-specific neural network (the q-net), whose weights are generated by a hypernetwork from the contextualized query embeddings. This design enables more expressive relevance estimation while preserving independent query and document encoding. In this work, we conduct a reproducibility study of the Hypencoder and extend the original analysis in three directions. Our reproduction confirms that the Hypencoder outperforms a similarly trained bi-encoder baseline on in-domain and out-of-domain benchmarks, and that the proposed efficient search algorithm substantially reduces query latency with minimal performance loss. On hard retrieval tasks, we find partial support: the Hypencoder outperforms the baseline on DL-Hard and FollowIR, but not on TREC TOT, where checkpoint incompatibility and fine-tuning sensitivity complicate full verification. Beyond reproduction, we investigate three extensions: (i) integrating alternative pre-trained encoders into the Hypencoder framework, where we find that performance gains depend on the encoder and fine-tuning strategy; (ii) comparing query latency against a Faiss-based bi-encoder pipeline, revealing that standard bi-encoder retrieval remains faster under both exhaustive and efficient search settings; and (iii) evaluating adversarial robustness, where we find that the q-net's non-linear scoring does not provide a consistent robustness disadvantage over inner-product scoring. Our code is publicly available at https://github.com/arneeichholtz/Hypencoder-reprod.

Dimensionality reduction is critical for deploying dense retrieval systems at scale, yet mainstream post-hoc methods face a fundamental trade-off: principal component analysis (PCA) preserves dominant variance but underutilizes representational capacity, while whitening enforces isotropy at the cost of amplifying noise in the heavy-tailed eigenspectrum of retrieval embeddings. Intermediate spectral scaling methods unify these extremes by reweighting dimensions with a power coefficient γ, but treat γ as a fixed hyperparameter that requires task-specific tuning. We show that the optimal scaling strength γ is not a global constant: it varies systematically with target dimensionality k and is governed by the signal-to-noise ratio (SNR) of the retained subspace. Based on this insight, we propose Spectral Tempering (SpecTemp), a learning-free method that derives an adaptive γ(k) directly from the corpus eigenspectrum using local SNR analysis and knee-point normalization, requiring no labeled data or validation-based search. Extensive experiments demonstrate that Spectral Tempering consistently achieves near-oracle performance relative to grid-searched γ*(k) while remaining fully learning-free and model-agnostic. Our code is publicly available at https://github.com/liyongkang123/SpecTemp.

Generative retrieval (GR) ranks documents by autoregressively generating document identifiers. Because many GR methods rely on trie-constrained beam search, they are vulnerable to early pruning of relevant prefixes under finite-beam decoding. Planning Ahead in Generative Retrieval (PAG) mitigates this failure mode by using simultaneous decoding to compute a document-level look-ahead prior that guides subsequent sequential decoding. We reproduce PAG at inference time and stress-test its decoding behavior. Using the authors' released checkpoint and identifier/trie artifacts under the reported decoding setup, we reproduce the main effectiveness results on MS~MARCO Dev and TREC-DL 2019/2020, and corroborate the reported beam-size--latency trade-off in our hardware setting. Beyond reproduction, we introduce plan drift diagnostics that quantify how intent-preserving query variations, including misspellings, reordering, synonym substitutions, paraphrases, and naturality shifts, alter the planner's top-n candidate set and highest-weight planner tokens, and how these changes affect guided decoding. We find that PAG's planning signal is brittle under lexical surface-form variation: intent-preserving typos can trigger plan collapse, where the planned candidate pool shifts enough that the look-ahead bonus provides little useful guidance, effectively reverting decoding toward weaker unguided search. We further evaluate fixed-index cross-lingual robustness using non-English mMARCO queries against an English index, and assess query-side mitigation strategies that require no re-indexing; query translation provides the strongest recovery in our setting. Overall, our results confirm PAG's reported effectiveness and the benefit of planning-guided decoding under the released inference setup, while showing that these gains depend on the stability of the planning signal under realistic query variation and query--document mismatch. Code available at https://github.com/kidist-amde/lost-in-decoding.

Text embeddings are fundamental to many natural language processing (NLP) tasks, extensively applied in domains such as recommendation systems and information retrieval (IR).Traditionally, transmitting embeddings instead of raw text has been seen as privacy-preserving.However, recent methods such as Vec2Text challenge this assumption by demonstrating that controlled decoding can successfully reconstruct original texts from black-box embeddings.The unexpectedly strong results reported by Vec2Text motivated us to conduct further verification, particularly considering the typically non-intuitive and opaque structure of highdimensional embedding spaces.In this work, we reproduce the Vec2Text framework and evaluate it from two perspectives: (1) validating the original claims, and (2) extending the study through targeted experiments.First, we successfully replicate the original key results in both in-domain and out-of-domain settings, with only minor discrepancies arising due to missing artifacts, such as model checkpoints and dataset splits.Furthermore, we extend the study by conducting a parameter sensitivity analysis, evaluating the feasibility of reconstructing sensitive inputs (e.g., passwords), and exploring embedding quantization as a lightweight privacy defense.Our results show that Vec2Text is effective under ideal conditions, capable of reconstructing even password-like sequences that lack clear semantics.However, we identify key limitations, including its sensitivity to input sequence length.We also find that Gaussian noise and quantization techniques can mitigate the privacy risks posed by Vec2Text, with quantization offering a simpler and more widely applicable solution.Our findings emphasize the need for caution in using text embeddings and highlight the importance of further research into robust defense mechanisms for NLP systems.Our code and experiment results are available at https://github.com/dqmis/vec2text-repro.

This paper concerns corpus poisoning attacks in dense information retrieval, where an adversary attempts to compromise the ranking performance of a search algorithm by injecting a small number of maliciously generated documents into the corpus.Our work addresses two limitations in the current literature.First, attacks that perform adversarial gradient-based word substitution search do so in the discrete lexical space, while retrieval itself happens in the continuous embedding space.We thus propose an optimization method that operates in the embedding space directly.Specifically, we train a perturbation model with the objective of maintaining the geometric distance between the original and adversarial document embeddings, while also maximizing the token-level dissimilarity between the original and adversarial documents.Second, it is common for related work to have a strong assumption that the adversary has prior knowledge about the queries.In this paper, we focus on a more challenging variant of the problem where the adversary assumes no prior knowledge about the query distribution (hence, unsupervised).Our core contribution is an adversarial corpus attack that is fast and effective.We present comprehensive experimental results on both in-and out-of-domain datasets, focusing on two related tasks: a top-1 attack and a corpus poisoning attack.We consider attacks under both a white-box and a black-box setting.Notably, our method can generate successful adversarial examples in under two minutes per target document; four times faster compared to the fastest gradientbased word substitution methods in the literature with the same hardware.Furthermore, our adversarial generation method generates text that is more likely to occur under the distribution of natural text (low perplexity), and is therefore more difficult to detect.

Friend recommendation is an important real-world application in Location-based Social Networks (LBSN), helping users discover potential friends and enhance their overall happiness. LBSN mainly comprises two distinct data structures: spatio-temporal data for human mobility and graph data for social networks. These two data structures make it challenging to model the complex relationships between them, which are essential for comprehensively understanding users’ lives. Previous studies have either modeled user trajectories and social networks separately or used classical simple graph-based methods, where a simple edge links only two nodes, failing to capture the multiple relationships inherent in LBSN. Furthermore, most studies have relied on Euclidean space to train their graph models, which could result in significant distortion because of tree-like social network data structure. To address these limitations, we propose a novel heterogeneous LBSN hypergraph that represents user check-in records and continuous trajectories—comprising multiple Points of Interest (POI)—as hyperedges, enabling the representation of complex spatio-temporal relationships. This approach enables us to link multiple nodes of different types by hyperedges and use hyperbolic spaces to create more efficient graph representations. Additionally, we devise a new type-specific attention mechanism for our Heterogeneous Hyperbolic Hypergraph Neural Network (H 3 GNN), which is end-to-end trainable and employs supervised contrastive learning to learn hypergraph node embeddings for the subsequent friend recommendation task with the help of hyperbolic space. Finally, our model H 3 GNN achieves better results than existing methods on six real-world city datasets, and our ablation studies demonstrate the effectiveness of each component. Additionally, our experiments indicate that H 3 GNN requires less data storage and training time compared to previous methods.

Friend recommendation from human mobility is a vital real-world application of location-based social networks (LBSN). It is necessary to recognize patterns from human mobility to assist friend recommendation because previous works have shown complex relations between them. However, most of previous works either modelled social networks and user trajectories separately, or only used classical simple graph-based methods with an edge linking two nodes that cannot fully model the complex data structure of LBSN. Inspired by the fact that hyperedges can connect multiple nodes of different types, we model user trajectories and check-in records as hyperedges in a novel heterogeneous LBSN hypergraph to represent complex spatio-temporal information. And then, we design a type-specific attention mechanism for an end-to-end trainable heterogeneous hypergraph neural network (HHGNN) with supervised contrastive learning, which can learn hypergraph node embedding for the next friend recommendation task. At last, our model HHGNN outperforms the state-of-the-art methods on four real-world city datasets, while ablation studies also confirm the effectiveness of each model part.