arXiv subjects

Jelena Mitrovic

Publications and source records attributed to Jelena Mitrovic.

13 recordsLinked to original sources

arXiv2026

The Compaction Cliff in Long-Running AI Agent Memory

A safety rule and an episodic log compete for the same tokens in an AI agent's context. When the budget overflows, both are summarized at the same rate; only the rule needs exact wording to remain enforceable. On 20 production agent configurations, Claude Code's /compact prompt on Sonnet 4.6 preserves 53\% of safety rules after one compaction round and 10\% after five. We name this the Compaction Cliff. We address it with Knowledge Triage, a framework that classifies each line of an agent's knowledge base by type and routes each type through its own retention policy. Three deterministic operators implement this triage across the three context-management operations: TypeCompact rewrites items in place under per-type fidelity, TypeDecompose partitions a topic too large to compact safely, replicating in-scope safety rules across partitions, and TypeRetrieve fetches items from external storage with in-scope rules pinned ahead of relevance. On five public corpora, TypeCompact preserves 2--4$\times$ more safety rules than the strongest single-shot LLM compactor at every ratio, with 96\% recall over five rounds. TypeDecompose reaches 0\% locality violations against 93\% under uniform partitioning. TypeRetrieve reaches 100\% recall@50 against 73\% for the best single-shot LLM retriever. On three downstream behavioral benchmarks, we outperform the production Sonnet compactor on medical compliance (paired McNemar $p < 10^{-8}$ on preservation, $N = 200$), the full-policy and hierarchical baselines on retail task pass rate ($p < 0.01$, $N = 115$), and the hierarchical compaction on the airline domain ($p = 0.024$). We release AgentArtifactCorpus (396{,}934 agent configurations from 54{,}628 public GitHub repositories), the classifier, and the reference implementation.

cs.AI↗

arXiv2026

Metadata, Structure, or Strategy? A Decomposition of RAG Context Enrichment

Retrieval-augmented generation (RAG) systems increasingly enrich retrieved passages by attaching quality metadata, structuring them into explicit records, and adopting multi-hop retrieval strategies that accumulate evidence across steps. These changes assume that richer context yields better answers, yet existing evaluations cannot test this because they vary all three factors at once. We isolate each factor in a controlled experiment across six benchmarks, four models from three families, and five enrichment levels, totaling over 24,000 evaluated responses. The assumption does not hold. Most enrichment reduces accuracy. Models prompted to use confidence scores comply correctly yet produce worse answers, a gap between utilization and accuracy that no prior work has measured. What determines answer quality is not how much metadata the context carries but whether the model can act on it for the given task. When metadata and retrieval strategy are aligned with model capabilities, a smaller model outperforms a frontier model by 19 F1 points. These findings motivate a processability hierarchy that predicts, from pre-training properties alone, which metadata a model can productively use, reframing RAG design as a question of model-context alignment rather than metadata accumulation.

cs.IR↗

arXiv2026

As We May Search

The sensitive information in personal documents, legal files, and medical records is among the most valuable things to search, yet current retrieval-augmented generation systems still require sending content to remote servers. We propose local-first IR, a design philosophy where indexes, models, and inference reside on user devices, treating remote services as optional. This paper makes four contributions: (1) a framework organizing retrieval architectures along three dimensions: privacy and control, capability, and accessibility, (2) experiments on consumer hardware across five benchmarks, scaling from 1K to 1M documents with dense retrieval, BM25, and hybrid fusion. Dense retrieval keeps over 91% nDCG@10 up to 100K documents, with approximate HNSW indexes extending this to 1M with only 2% quality loss; a 7B local language model reaches within 4 points of a cloud baseline on answer quality, (3) competing perspectives for and against local-first IR, informed by experimental evidence, and (4) a research agenda identifying open problems. The real tradeoff is scope rather than quality: what matters is what you can search, not how well you can search it.

cs.IR↗

arXiv2026

GradeLegal: Automated Grading for German Legal Cases

Grading German legal exam solutions faces growing volumes and a shortage of qualified graders, delaying feedback and creating a bottleneck. At the same time, it is a high-stakes expert task, since state exam grades strongly influence career outcomes in Germany. Despite this practical relevance, literature lacks systematic studies on effective methods for grading legal exams. To address this gap, we investigate whether large language models (LLMs) can support the automated grading of German legal case solutions in criminal and public law, thereby enabling scalable feedback and student self-testing. We present a systematic evaluation of 27 proprietary and open-source LLMs, benchmarking prompting strategies that incrementally add task-related information, such as a sample solution and a grading rubric. Using quadratic weighted kappa (QWK), reasoning-oriented LLMs can approximate expert grading in public law when given a sample solution and a grading rubric (up to 0.91), compared to 0.60 in criminal law, suggesting a harder grading task in criminal law. Beyond single-model grading, ensembling improves agreement by up to 0.15 over its best member and can offer an alternative to stronger closed-source single models. In addition, our findings suggest that effective prompt design and model selection are necessary for reliable LLM-based grading of legal exams.

cs.CL↗

arXiv2026

NuggetIndex: Governed Atomic Retrieval for Maintainable RAG

Retrieval-augmented generation (RAG) systems are frequently evaluated via fact-based metrics, yet standard implementations retrieve passages or static propositions. This unit mismatch between evaluation and retrieval objects hinders maintenance when corpora evolve and fails to capture superseded facts or source disagreements. We propose NuggetIndex, a retrieval system that stores atomic information units as managed records, so called nuggets. Each record maintains links to evidence, a temporal validity interval, and a lifecycle state. By filtering invalid or deprecated nuggets prior to ranking, the system prevents the inclusion of outdated information. We evaluate the approach using a nuggetized MS MARCO subset, a temporal Wikipedia QA dataset, and a multi-hop QA task. Against passage and unmanaged proposition retrieval baselines, NuggetIndex improves nugget recall by 42%, increases temporal correctness by 9 percentage points without the recall collapse observed in time-filtered baselines, and reduces conflict rates by 55%. The compact nugget format reduces generator input length by 64% while enabling lightweight index structures suitable for browser-based and resource-constrained deployment. We release our implementation, datasets, and evaluation scripts

cs.IR↗

arXiv2026

AgentSim: A Platform for Verifiable Agent-Trace Simulation

Training trustworthy agentic LLMs requires data that shows the grounded reasoning process, not just the final answer. Existing datasets fall short: question-answering data is outcome-only, chain-of-thought data is not tied to specific documents, and web-agent datasets track interface actions rather than the core retrieval and synthesis steps of a RAG workflow. We introduce AgentSim, an open-source platform for simulating RAG agents. It generates verifiable, stepwise traces of agent reasoning over any document collection. AgentSim uses a policy to ensure the agent widely explores the document set. It combines a multi-model validation pipeline with an active human-in-the-loop process. This approach focuses human effort on difficult steps where models disagree. Using AgentSim, we construct and release the Agent-Trace Corpus (ATC), a large collection of grounded reasoning trajectories spanning three established IR benchmarks. We make three contributions: (1) the AgentSim platform with two mechanisms, Corpus-Aware Seeding and Active Validation, that improve trace diversity and quality; (2) the Agent-Trace Corpus (ATC), over 103,000 verifiable reasoning steps spanning three IR benchmarks, with 100% grounding rate on substantive answers; and (3) a comparative behavioral analysis revealing systematic differences in how state-of-the-art models approach information seeking. Platform, toolkit, and corpus are publicly available.

cs.IR↗

arXiv2026

Form Without Function: Agent Social Behavior in the Moltbook Network

Moltbook is a social network where every participant is an AI agent. We analyze 1,312,238 posts, 6.7~million comments, and over 120,000 agent profiles across 5,400 communities, collected over 40 days (January 27 to March 9, 2026). We evaluate the platform through three layers. At the interaction layer, 91.4% of post authors never return to their own threads, 85.6% of conversations are flat (no reply ever receives a reply), the median time-to-first-comment is 55 seconds, and 97.3% of comments receive zero upvotes. Interaction reciprocity is 3.3%, compared to 22-60% on human platforms. An argumentation analysis finds that 64.6% of comment-to-post relations carry no argumentative connection. At the content layer, 97.9% of agents never post in a community matching their bio, 92.5% of communities contain every topic in roughly equal proportions, and over 80% of shared URLs point to the platform's own infrastructure. At the instruction layer, we use 41 Wayback Machine snapshots to identify six instruction changes during the observation window. Hard constraints (rate limit, content filters) produce immediate behavioral shifts. Soft guidance (``upvote good posts'', ``stay on topic'') is ignored until it becomes an explicit step in the executable checklist. The platform also poses technological risks. We document credential leaks (API keys, JWT tokens), 12,470 unique Ethereum addresses with 3,529 confirmed transaction histories, and attack discourse ranging from template-based SSH brute-forcing to multi-agent offensive security architectures. These persist unmoderated because the quality-filtering mechanisms are themselves non-functional. Moltbook is a socio-technical system where the technical layer responds to changes, but the social layer largely fails to emerge. The form of social media is reproduced in full. The function is absent.

cs.SI↗

arXiv2026

Behind the Prompt: The Agent-User Problem in Information Retrieval

User models in information retrieval rest on a foundational assumption that observed behavior reveals intent. This assumption collapses when the user is an AI agent privately configured by a human operator. For any action an agent takes, a hidden instruction could have produced identical output - making intent non-identifiable at the individual level. This is not a detection problem awaiting better tools; it is a structural property of any system where humans configure agents behind closed doors. We investigate the agent-user problem through a large-scale corpus from an agent-native social platform: 370K posts from 47K agents across 4K communities. Our findings are threefold: (1) individual agent actions cannot be classified as autonomous or operator-directed from observables; (2) population-level platform signals still separate agents into meaningful quality tiers, but a click model trained on agent interactions degrades steadily (-8.5% AUC) as lower-quality agents enter training data; (3) cross-community capability references spread endemically ($R_0$ 1.26-3.53) and resist suppression even under aggressive modeled intervention. For retrieval systems, the question is no longer whether agent users will arrive, but whether models built on human-intent assumptions will survive their presence.

cs.IR↗

arXiv2026

OwlerLite: Scope- and Freshness-Aware Web Retrieval for LLM Assistants

Browser-based language models often use retrieval-augmented generation (RAG) but typically rely on fixed, outdated indices that give users no control over which sources are consulted. This can lead to answers that mix trusted and untrusted content or draw on stale information. We present OwlerLite, a browser-based RAG system that makes user-defined scopes and data freshness central to retrieval. Users define reusable scopes-sets of web pages or sources-and select them when querying. A freshness-aware crawler monitors live pages, uses a semantic change detector to identify meaningful updates, and selectively re-indexes changed content. OwlerLite integrates text relevance, scope choice, and recency into a unified retrieval model. Implemented as a browser extension, it represents a step toward more controllable and trustworthy web assistants.

cs.HC↗

arXiv2025

Compressed Concatenation of Small Embedding Models

Embedding models are central to dense retrieval, semantic search, and recommendation systems, but their size often makes them impractical to deploy in resource-constrained environments such as browsers or edge devices. While smaller embedding models offer practical advantages, they typically underperform compared to their larger counterparts. To bridge this gap, we demonstrate that concatenating the raw embedding vectors of multiple small models can outperform a single larger baseline on standard retrieval benchmarks. To overcome the resulting high dimensionality of naive concatenation, we introduce a lightweight unified decoder trained with a Matryoshka Representation Learning (MRL) loss. This decoder maps the high-dimensional joint representation to a low-dimensional space, preserving most of the original performance without fine-tuning the base models. We also show that while concatenating more base models yields diminishing gains, the robustness of the decoder's representation under compression and quantization improves. Our experiments show that, on a subset of MTEB retrieval tasks, our concat-encode-quantize pipeline recovers 89\% of the original performance with a 48x compression factor when the pipeline is applied to a concatenation of four small embedding models.

cs.LG↗

arXiv2025

Krony-PT: GPT2 compressed with Kronecker Products

We introduce Krony-PT, a compression technique for GPT-2 based on Kronecker products. We specifically target the feed-forward weights of each transformer block, and systematically compress the feed-forward layer matrices to various degrees. We introduce a modified Van Loan decomposition to initialize new Kronecker factors, and also propose a new pruning-based initialization technique. Our method compresses the original 124M-parameter GPT-2 to various smaller models, ranging from 80M to 96M. Our 81M model variant outperforms DistilGPT2 on next-token prediction across all standard language modeling datasets, and shows competitive or comparable performance with significantly larger Kronecker-based compressions of GPT-2.

cs.LG↗

arXiv2024

SUDS: A Strategy for Unsupervised Drift Sampling

Supervised machine learning often encounters concept drift, where the data distribution changes over time, degrading model performance. Existing drift detection methods focus on identifying these shifts but often overlook the challenge of acquiring labeled data for model retraining after a shift occurs. We present the Strategy for Drift Sampling (SUDS), a novel method that selects homogeneous samples for retraining using existing drift detection algorithms, thereby enhancing model adaptability to evolving data. SUDS seamlessly integrates with current drift detection techniques. We also introduce the Harmonized Annotated Data Accuracy Metric (HADAM), a metric that evaluates classifier performance in relation to the quantity of annotated data required to achieve the stated performance, thereby taking into account the difficulty of acquiring labeled data. Our contributions are twofold: SUDS combines drift detection with strategic sampling to improve the retraining process, and HADAM provides a metric that balances classifier performance with the amount of labeled data, ensuring efficient resource utilization. Empirical results demonstrate the efficacy of SUDS in optimizing labeled data use in dynamic environments, significantly improving the performance of machine learning applications in real-world scenarios. Our code is open source and available at https://github.com/cfellicious/SUDS/

cs.LG↗

arXiv2024

Beyond Benchmarks: Evaluating Embedding Model Similarity for Retrieval Augmented Generation Systems

The choice of embedding model is a crucial step in the design of Retrieval Augmented Generation (RAG) systems. Given the sheer volume of available options, identifying clusters of similar models streamlines this model selection process. Relying solely on benchmark performance scores only allows for a weak assessment of model similarity. Thus, in this study, we evaluate the similarity of embedding models within the context of RAG systems. Our assessment is two-fold: We use Centered Kernel Alignment to compare embeddings on a pair-wise level. Additionally, as it is especially pertinent to RAG systems, we evaluate the similarity of retrieval results between these models using Jaccard and rank similarity. We compare different families of embedding models, including proprietary ones, across five datasets from the popular Benchmark Information Retrieval (BEIR). Through our experiments we identify clusters of models corresponding to model families, but interestingly, also some inter-family clusters. Furthermore, our analysis of top-k retrieval similarity reveals high-variance at low k values. We also identify possible open-source alternatives to proprietary models, with Mistral exhibiting the highest similarity to OpenAI models.

cs.IR↗