Information retrieval (IR) systems have traditionally been designed and trained for human users, with learning-to-rank methods relying heavily on large-scale human interaction logs such as clicks and dwell time. With the rapid emergence of large language model (LLM) powered search agents, however, retrieval is increasingly consumed by agents rather than human beings, and is embedded as a core component within multi-turn reasoning and action loops. In this setting, retrieval models trained under human-centric assumptions can be mismatched with the way agents issue intermediate queries and consume results. In this work, we argue that retrieval models for agentic search should be trained directly from agent interaction data. We study learning to retrieve from agent trajectories as a trajectory-supervised training setting, where supervision is derived from multi-step agent interactions. Through a systematic analysis of search agent trajectories, we identify key behavioral signals that reveal document utility, including browsing actions, unbrowsed rejections, and post-browse reasoning traces. Guided by these insights, we propose LRAT, a simple yet effective framework that mines high-quality retrieval supervision from agent trajectories and incorporates relevance intensity through weighted optimization. To instantiate this setting at scale, we deploy the Tongyi-DeepResearch-30B model on 10K InfoSeekQA queries with four retrievers, collecting 26,482 agent trajectories and constructing 91,713 training pairs. Extensive experiments on both in-domain and out-of-domain deep research benchmarks demonstrate that retrievers trained with LRAT consistently improve evidence recall, end-to-end task success, and execution efficiency across diverse agent architectures and scales. Our results highlight agent trajectories as a practical and scalable supervision source for retrieval in agentic search.

Text embeddings from PLM-based models enable a wide range of applications, yet their performance often degrades on longer texts.In this paper, we introduce a phenomenon we call Length Collapse, where embeddings of longer texts tend to cluster together.This clustering results in a distributional inconsistency between the embeddings of short and long texts.We further investigate how these differences contribute to the performance decline observed with longer texts across various downstream tasks.Through a rigorous theoretical analysis of the self-attention mechanism, which acts as a low-pass filter in PLM-based models, we demonstrate that as text length increases, the strength of low-pass filtering intensifies, causing embeddings to retain more low-frequency components.As a result, input token features become more similar, leading to clustering and ultimately the collapse of embeddings for longer texts.To address this issue, we propose a simple method, TempScale, which mitigates the Length Collapse phenomenon.By narrowing the gap in low-pass filtering rates between long and short texts, TempScale ensures more consistent embeddings across different text lengths.This approach leads to performance improvements of 0.94% on MTEB and 1.10% on LongEmbed, which focuses specifically on long-context retrieval, providing strong evidence for the validity of our analysis.

Recent Graphical User Interface (GUI) agents replicate the R1-Zero paradigm, coupling online Reinforcement Learning (RL) with explicit chain-of-thought reasoning prior to object grounding and thereby achieving substantial performance gains. In this paper, we first conduct extensive analysis experiments of three key components of that training pipeline: input design, output evaluation, and policy update—each revealing distinct challenges arising from blindly applying general-purpose RL without adapting to GUI grounding tasks. Input design: Current templates encourage the model to generate chain-of-thought reasoning, but longer chains unexpectedly lead to worse grounding performance. Output evaluation: Reward functions based on hit signals or box area allow models to exploit box size, leading to reward hacking and poor localization quality. Policy update: Online RL tends to overfit easy examples due to biases in length and sample difficulty, leading to under-optimization on harder cases. To address these issues, we propose three targeted solutions. First, we adopt a $\textbf{Fast Thinking Template}$ that encourages direct answer generation, reducing excessive reasoning during training. Second, we incorporate a box size constraint into the reward function to mitigate reward hacking. Third, we revise the RL objective by adjusting length normalization and adding a difficulty-aware scaling factor, enabling better optimization on hard samples. Our $\textbf{GUI-G1-3B}$, trained on 17K public samples with Qwen2.5-VL-3B-Instruct, achieves $\textbf{90.3\%}$ accuracy on ScreenSpot and $\textbf{37.1\%}$ on ScreenSpot-Pro. This surpasses all prior models of similar size and even outperforms the larger UI-TARS-7B, establishing a new state-of-the-art in GUI agent grounding.

Recommender systems are essential for information access, allowing users to present their content for recommendation. With the rise of large language models (LLMs), AI-generated content (AIGC), primarily in the form of text, has become a central part of the content ecosystem. As AIGC becomes increasingly prevalent, it is important to understand how it affects the performance and dynamics of recommender systems. To this end, we construct an environment that incorporates AIGC to explore its short-term impact. The results from popular sequential recommendation models reveal that AIGC are ranked higher in the recommender system, reflecting the phenomenon of source. To further explore the long-term impact of AIGC, we introduce a feedback loop with realistic simulators. The results show that the model's preference for AIGC increases as the user clicks on AIGC rises and the model trains on simulated click data. This leads to two issues: In the short term, bias toward AIGC encourages LLM-based content creation, increasing AIGC content, and causing unfair traffic distribution. From a long-term perspective, our experiments also show that when AIGC dominates the content ecosystem after a feedback loop, it can lead to a decline in recommendation performance. To address these issues, we propose a debiasing method based on L1-loss optimization to maintain long-term content ecosystem balance. In a real-world environment with AIGC generated by mainstream LLMs, our method ensures a balance between AIGC and human-generated content in the ecosystem. The code and dataset are available at https://github.com/Yuqi-Zhou/Rec_SourceBias.

Recent studies have revealed a phenomenon known as source bias, where PLM-based retrievers assign higher relevance scores to LLM-generated content despite its semantic quality being comparable to human-written content. As LLMs rapidly advance and become more widely used, effectively counteracting source bias is crucial for the sustainable development of the information retrieval (IR) ecosystem. Existing methods primarily attempt to address source bias from the retriever side, adopting a "passive defense" approach that intervenes only after biased content has entered the retrieval pipeline. These solutions are limited by frequent retriever updates in industrial applications, high recurring costs, and their inability to address the root cause of source bias.

Recently, the emergence of large language models (LLMs) has revolutionized the paradigm of information retrieval (IR) applications, especially in web search, by generating vast amounts of human-like texts on the Internet. As a result, IR systems in the LLM era are facing a new challenge: the indexed documents are now not only written by human beings but also automatically generated by the LLMs. How these LLM-generated documents influence the IR systems is a pressing and still unexplored question. In this work, we conduct a quantitative evaluation of IR models in scenarios where both human-written and LLM-generated texts are involved. Surprisingly, our findings indicate that neural retrieval models tend to rank LLM-generated documents higher. We refer to this category of biases in neural retrievers towards the LLM-generated content as the source bias. Moreover, we discover that this bias is not confined to the first-stage neural retrievers, but extends to the second-stage neural re-rankers. Then, in-depth analyses from the perspective of text compression indicate that LLM-generated texts exhibit more focused semantics with less noise, making it easier for neural retrieval models to semantic match. To mitigate the source bias, we also propose a plug-and-play debiased constraint for the optimization objective, and experimental results show its effectiveness. Finally, we discuss the potential severe concerns stemming from the observed source bias and hope our findings can serve as a critical wake-up call to the IR community and beyond. To facilitate future explorations of IR in the LLM era, the constructed two new benchmarks are available at https://github.com/KID-22/Source-Bias.

In online industrial advertising systems, conversion actions (e.g., purchases or downloads) often occur significantly delayed, even up to several days or weeks after the user clicks. This phenomenon leads to the crucial challenge calleddelayed feedback problem in streaming CVR prediction, that is, the online systems cannot receive the true label of conversions immediately for continuous training. To mitigate the delayed feedback problem, recent state-of-the-art methods often apply sample duplicate mechanisms to introduce early certain conversion information. Nevertheless, these works have overlooked a crucial issue of rapid shifts in data distribution and considered both the newly observed data and duplicated early data together, resulting in biases in both distributions. In this work, we propose a Dually enhanced Delayed Feedback Model (DDFM), which tackles the above issues by treating the newly observed data and duplicated early data separately. DDFM consists of dual unbiased CVR estimators that share the same form but utilize different latent variables as weights: one for the newly observed data and the other for the duplicated early data. To avoid high variance, we adopt an addition-only formula for these latent variables, eliminating multiplication or division operations. Furthermore, we design a shared-bottom network that efficiently and jointly estimates the latent variables in DDFM. Theoretical analysis demonstrates the unbiasedness and convergence properties of DDFM. Extensive experiments on both public and industrial large-scale real-world datasets exhibit that our proposed DDFM consistently outperforms existing state-of-the-art methods.