Lost in the Middle: How Language Models Use Long Contexts
Open accessThe DOI matches its Crossref record; 48 of the 59 references it cites were found in a registry or an index; 15 of the 23 sources its claims cite were checked for support.
Abstract
While recent language models have the ability to take long contexts as input, relatively little is known about how well they use longer context. We analyze the performance of language models on two tasks that require identifying relevant information in their input contexts: multi-document question answering and key-value retrieval. We find that performance can degrade significantly when changing the position of relevant information, indicating that current language models do not robustly make use of information in long input contexts. In particular, we observe that performance is often highest when relevant information occurs at the beginning or end of the input context, and significantly degrades when models must access relevant information in the middle of long contexts, even for explicitly long-context models. Our analysis provides a better understanding of how language models use their input context and provides new evaluation protocols for future long-context language models.
Overview
CoChat Summary
TL;DR
Language models’ performance on tasks requiring access to long input contexts is strongly affected by where relevant information appears, with a U-shaped curve: best at the beginning or end and worst in the middle. Extended-context models are not necessarily better at using context, and reader performance saturates before retriever recall in open-domain QA.
Key findings
- In multi-document QA, performance is highest when the answer-containing document is at the start or end of the context and degrades in the middle; GPT-3.5-Turbo can drop by more than 20%, and in 20- and 30-document settings middle-position performance is lower than its closed-book performance of 56.1% (oracle: 88.3%).
- Extended-context models are not necessarily better at using input context: when the input fits both windows, GPT-3.5-Turbo and GPT-3.5-Turbo(16K) perform nearly identically as a function of relevant-information position.
- In synthetic key-value retrieval, Claude-1.3 and Claude-1.3(100K) do nearly perfectly, while other models struggle especially with 140 or 300 key-value pairs; query-aware contextualization gives GPT-3.5-Turbo(16K) perfect performance at 300 pairs, versus worst-case 45.6% without it.
- In open-domain QA, reader performance saturates long before retriever recall: using 50 instead of 20 retrieved documents only marginally improves performance (∼1.5% for GPT-3.5-Turbo and ∼1% for Claude-1.3).
Methods
- Controlled multi-document QA on NaturalQuestions-Open: 2,655 queries whose annotated long answer is a paragraph; each context has exactly one answer-containing Wikipedia passage plus k−1 Contriever/MS-MARCO distractor passages; the paper varies answer position and context length (10, 20, 30 documents) and measures accuracy.
- Synthetic key-value retrieval: inputs are serialized JSON objects with unique random UUID key-value pairs; models must return the value for a specified key; experiments use 75, 140, and 300 key-value pairs (500 examples each), varying key position and number of pairs, with accuracy as the metric.
- Models include open MPT-30B-Instruct and LongChat-13B(16K), closed GPT-3.5-Turbo, GPT-3.5-Turbo(16K), Claude-1.3, and Claude-1.3(100K), plus additional architecture/instruction experiments with Flan-T5-XXL, Flan-UL2, MPT-30B, Llama-2, and GPT-4; an open-domain QA case study uses a Contriever retriever-reader setup.
Limitations
- Stated: the paper uses greedy decoding and leaves exploration of other decoding methods to future work; the investigations into why models are not robust are described as preliminary; GPT-4 is evaluated only on a subset of 500 multi-document QA examples because full evaluation would cost upwards of $6000.
- Stated: NaturalQuestions-Open has ambiguity, so the authors run an unambiguous subset and also test random distractors and randomized distractor order, finding similar trends.
- Evident: the controlled multi-document QA setting always contains exactly one document that answers the question, unlike open-domain QA; the synthetic key-value task removes natural language semantics; results may not generalize to all tasks, models, or decoding settings.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
7 Valid · 0 Not valid · 5 Not checked · 13 Own results and interpretations
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Language models have become an important and flexible building block in a variety of user-facing language technologies, including conversational interfaces, search and summarization, and collaborative writing (Shuster et al., 2022;Thoppilan et al., 2022;Lee et al., 2022, inter alia).”
§ Lost in the Middle: How Language Models Use Long Contexts > 1 Introduction
What the sources say
- Partly supportedcites Shuster et al., 2022 matched to “BlenderBot 3: a deployed conversational agent that continually learns to responsibly engage” (2022) sourceExperimental — accuracy not yet measured
Section 3.1.1 (p. 4) describes a conversational agent built on a language model that generates internet search queries, supporting conversational interfaces and search as part of dialogue. Section 2 (p. 3) mentions summarization only as related work, and the passages do not establish collaborative writing as an application of this paper’s model. (LLM, conf=0.85)
- Supportedcites Thoppilan et al., 2022 matched to “LaMDA: Language Models for Dialog Applications” (2022) source
The Abstract (p. 1) identifies LaMDA as a family of language models specialized for dialog. In this sentence's list of applications, Thoppilan et al. supports the conversational-interface example; the other cited works need not each cover every example. (LLM, conf=0.92)
- Supportedcites Lee et al., 2022 matched to “CoAuthor: Designing a Human-AI Collaborative Writing Dataset for Exploring Language Model Capabilities” (2022) source
Lee et al. is cited alongside other works, and its contribution to the sentence is the collaborative-writing example. Section 4.2.2 describes writers receiving and selecting model-generated suggestions in a writing interface; Section 2.2.1 also describes language models as applicable to dialogue and summarization. The paper need not establish every example in the jointly cited list. (LLM, conf=0.89)
- Partly supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“These observations complement prior work, which found that non-instruction fine-tuned language models are biased towards recent tokens (i.e., the end of the input context; Khandelwal et al., 2018;Press et al., 2021).”
§ Lost in the Middle: How Language Models Use Long Contexts > 4.3 Effect of Instruction Fine-Tuning
What the sources say
- Partly supportedcites Khandelwal et al., 2018 matched to “Sharp nearby, fuzzy far away: How neural language models use context” (2018) sourceExperimental — accuracy not yet measured
The Introduction and Section 5 report that the studied LSTM is much more sensitive to recent word order than to distant word order, supporting a recency effect. The paper studies a standard LSTM on two datasets, however, rather than establishing this tendency across non-instruction-fine-tuned language models generally; it also finds that distant words remain relevant. (LLM, conf=0.82)
- Partly supportedcites Press et al., 2021 matched to “Shortformer: Better language modeling using shorter inputs” (2021) sourceExperimental — accuracy not yet measured
Section 3.1 (page not available in our copy) finds that very short contexts hurt performance, while increasing subsequence length from 1,024 to 3,072 tokens gives negligible improvement. That is consistent with limited benefit from more distant context, but the paper does not establish that the models are biased toward recent tokens within an input context. (LLM, conf=0.78)
- Partly supported
The sources with a verdict support this claim; the rest weren’t checked yet, have no full text we could read, or their verdicts aren’t rated yet.
In the paper
“We instantiate this task with data from NaturalQuestions-Open (Lee et al., 2019;Kwiatkowski et al., 2019), which contains historical queries issued to the Google search engine, coupled with human-annotated answers extracted from Wikipedia.”
§ Lost in the Middle: How Language Models Use Long Contexts > 2.1 Experimental Setup
What the sources say
- Supportedcites Lee et al., 2019 matched to “Latent retrieval for weakly supervised open domain question answering” (2019) source
Section 7.1 says the paper creates an open version of Natural Questions and uses it for training and evaluation. The Introduction describes the task as answering questions by retrieving evidence from an open corpus, supporting the cited use of NaturalQuestions-Open for this task. (LLM, conf=0.91)
- Not checked — no full text of the sourcecites Kwiatkowski et al., 2019 matched to “Natural Questions: A benchmark for question answering research” (2019) source
- Supported
Every source the paper cites for this claim was checked and supports it.
In the paper
“Following Kandpal et al. (2022) and Mallen et al. (2023), we use accuracy as our primary evaluation metric, judging whether any of the cor- rect answers (as taken from the NaturalQuestions annotations) appear in the predicted output.”
§ Lost in the Middle: How Language Models Use Long Contexts > 2.1 Experimental Setup
What the sources say
- Supportedcites Kandpal et al. (2022) matched to “Large Language Models Struggle to Learn Long-Tail Knowledge” (2022) source
Section 3 (p. 4) says the paper evaluates answers using exact match against a ground-truth answer set, and Section 3.1 (p. 4) reports QA accuracies as its evaluation results. This supports citing Kandpal et al. for using accuracy to evaluate question answering. (LLM, conf=0.90)
- Supportedcites Mallen et al. (2023) matched to “When not to trust language models: Investigating effectiveness of parametric and non-parametric memories” (2023) source
Section 3.1 (‘Focus and Task’) explicitly names accuracy as the metric and says a prediction is correct when it contains an exact match to any gold answer. This supports the evaluation approach attributed to Mallen et al.; the NaturalQuestions annotations describe the citing authors’ answer source. (LLM, conf=0.94)
- Supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“or when language models are augmented with external information (relevant documents from a search engine, database query results, etc.; Petroni et al., 2020;Ram et al., 2023;Shi et al., 2023;Mallen et al., 2023;Schick et al., 2023, inter alia).”
§ Lost in the Middle: How Language Models Use Long Contexts > 1 Introduction
What the sources say
- Partly supportedcites Petroni et al., 2020 matched to “How Context Affects Language Models' Factual Predictions” (2020) sourceExperimental — accuracy not yet measured
The Abstract (p. 1) says the paper integrates information from a retrieval system with a pre-trained language model, and §3.4.2 (p. 5) specifies a retrieved paragraph as context. This supports augmentation with retrieved documents, but our copy does not show language models being augmented with database query results. (LLM, conf=0.84)
- Partly supportedcites Ram et al., 2023 matched to “In-Context Retrieval-Augmented Language Models” (2023) sourceExperimental — accuracy not yet measured
The Introduction describes augmenting language models with relevant documents retrieved from an external knowledge source, and Section 5 describes using a BM25 retriever to find those documents. The passages do not establish database query results as another type of external information used by this paper. (LLM, conf=0.82)
- Partly supportedcites Shi et al., 2023 matched to “REPLUG: Retrieval-augmented black-box language models” (2023) sourceExperimental — accuracy not yet measured
Section 3.1 (page unavailable) describes retrieving relevant documents from a corpus, and the Conclusion says REPLUG uses those documents to augment a language model. The paper supports the document-retrieval example, but our copy does not describe augmenting models with database query results or documents from a search engine specifically. (LLM, conf=0.82)
- Partly supported
20 more claims · 2 with their cited sources checked
Create a free account to see all 25 claims and every source verdict.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“In vanilla multi-document or NIAH-based benchmarks (Kamradt, 2023; Liu et al., 2024), models perform consistently well when the needle (or gold document) appears at the very beginning or end of the context window, with minimal impact from context length.”
From NoLiMa: Long-Context Evaluation Beyond Literal Matching (2025) · cites this paper as Liu et al., 2024
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Even within a single long sequence, attention effectiveness degrades with distance due to attention dilution , positional encoding limitations, and token interference, leading to the well-known 'lost-in-the-middle' and context-decay phenomena (Liu et al., 2024; Press et al., 2021a).”
From MAGMA: A Multi-Graph based Agentic Memory Architecture for AI Agents (2026) · cites this paper as Liu et al., 2024
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Current large language models (LLMs) exhibit significant deficiencies in episodic memory tasks including encoding, storing, and retrieving specific information from temporally dependent events over a long period of time.”
From ARTEM: Enhancing Large Language Model Agents with Spatial-Temporal Episodic Memory (2026) · cites this paper as Liu et al. 2024
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Recent approaches to handle memory tasks in LLMs, such as in-context learning, retrieval-augmented generation (RAG), and fine-tuning, may resolve the long-term retention issues, but are still inadequate to handle tasks requiring chronological awareness of the stored information.”
From ARTEM: Enhancing Large Language Model Agents with Spatial-Temporal Episodic Memory (2026) · cites this paper as Liu et al. 2024
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“If not ranked at the top, the key information may be neglected by the generative module, resulting in incomplete or inaccurate responses to the user queries24.”
From Leveraging long context in retrieval augmented language models for medical question answering (2025) · cites this paper as [24]
Experimental — accuracy not yet measured
86 more claims about this paper
Create a free account to see every claim other papers make about this one.
Citations
Cites
Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, L., & Polosukhin, I. (2017) · arXiv · doi:10.48550/arxiv.1706.03762 · Cited by 120,404
Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer
Raffel, C., Shazeer, N., Roberts, A., Lee, K., Narang, S., Matena, M., Zhou, Y., Li, W., & Liu, P. J. (2019) · arXiv · doi:10.48550/arxiv.1910.10683 · Cited by 10,929
LLaMA: Open and Efficient Foundation Language Models
Touvron, H., Lavril, T., Izacard, G., Martinet, X., Lachaux, M.-A., Lacroix, T., Rozière, B., Goyal, N., Hambro, E., Azhar, F., et al. (2023) · arXiv · doi:10.48550/arxiv.2302.13971 · Cited by 4,602
Transformer-XL: Attentive language models beyond a fixed-length context
Dai, Z., Yang, Z., Yang, Y., Carbonell, J., Le, Q., & Salakhutdinov, R. (2019) · Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics · doi:10.18653/v1/p19-1285 · Cited by 3,169 · cited 2×
“Recent improvements in hardware (e.g., faster GPUs with more memory) and algorithms (Dai et al., 2019; Dao et al., 2022; Poli et al., 2023; Rubin and Berant, 2023, inter alia ) have resulted in language models with larger context windows (e.g., 4096, 32K, and even 100K tokens), but it remains…”
BackgroundLlama 2: Open Foundation and Fine-Tuned Chat Models
Touvron, H., Martin, L., Stone, K., Albert, P., Almahairi, A., Babaei, Y., Bashlykov, N., Batra, S., Bhargava, P., Bhosale, S., et al. (2023) · arXiv · doi:10.48550/arxiv.2307.09288 · Cited by 2,604
Longformer: The Long-Document Transformer
Beltagy, I., Peters, M. E., & Cohan, A. (2020) · arXiv · doi:10.48550/arxiv.2004.05150 · Cited by 2,281
Leveraging Passage Retrieval with Generative Models for Open Domain Question Answering
Izacard, G., & Grave, E. (2020) · Conference of the European Chapter of the Association for Computational Linguistics · doi:10.48550/arxiv.2007.01282 · Cited by 2,281
Natural Questions: A benchmark for question answering research
Kwiatkowski, T., Palomaki, J., Redfield, O., Collins, M., Parikh, A., Alberti, C., Epstein, D., Polosukhin, I., Devlin, J., Lee, K., et al. (2019) · Transactions of the Association for Computational Linguistics · doi:10.1162/tacl_a_00276 · Cited by 2,069 · cited 1×
“We instantiate this task with data from NaturalQuestions-Open (Lee et al., 2019; Kwiatkowski et al., 2019), which contains historical queries issued to the Google search engine, coupled with human-annotated answers extracted from Wikipedia.”
MethodThe serial position effect of free recall.
Murdock, B. B. (1962) · Journal of Experimental Psychology · doi:10.1037/h0045106 · Cited by 1,717 · cited 1×
“The U-shaped curve we observe in this work has a connection in psychology known as the serial-position effect (Ebbinghaus, 1913; Murdock Jr, 1962), that states that in free-association recall of elements from a list, humans tend to best re-member the first and last elements of the list.”
MethodMemory: A contribution to experimental psychology.
Ebbinghaus, H. (1913) · doi:10.1037/10011-000 · Cited by 1,645 · cited 1×
“The U-shaped curve we observe in this work has a connection in psychology known as the serial-position effect (Ebbinghaus, 1913; Murdock Jr, 1962), that states that in free-association recall of elements from a list, humans tend to best re-member the first and last elements of the list.”
Method
Research works printed but not linked: 1
- Sharan, 2018, Prediction with a short memory, 10.18653/v1/K18-3013 · DOI
Cited by
Items 1,001–1,100 of 1,598
Designing Prompts for Shared-Experience-Focused Utterance Generation in LLM-Based Dialogue Systems
Fukushima, R., & Hashimoto, M. (2025) · Journal of Machine Intelligence and Data Science · doi:10.11159/jmids.2025.009
SES-RAG: An Adaptive Framework with ModernBERT Segmentation and Generative-Alignment Expansion
Gao, C. S., He, H., & Zhang, Q. (2026) · Applied Sciences · doi:10.3390/app16104849
Brevik, A., Jerpseth, H., & Lafontan, S. R. (2026) · Frontiers in Education · doi:10.3389/feduc.2026.1904452
Wang, Y., Jia, R., Zhang, W., Lü, W., Jiang, Y., Zhu, J., Wu, S., & Chen, T. (2026) · International Conference on Computer Supported Cooperative Work in Design (CSCWD) · doi:10.1109/cscwd68734.2026.11581682
“However, this approach faces challenges such as truncation of key information in long documents and insufficient extraction of salient content, often resulting in hallucinations [15].”
Tarannum, A., Mohammed, M. A., Mandalawi, S. A., Çakmak, M. C., & Talburt, J. R. (2026) · Knowledge · doi:10.3390/knowledge6030023
“Limiting each worker’s scope reduces prompt complexity and helps mitigate cognitive overload [12,47].”
InfluentialAnthony, C. R. O., & Peiris, S. M. (2026) · Moratuwa Engineering Research Conference (MERCon) · doi:10.1109/mercon71835.2026.11691559
“In addition, Liu et al. [3] states that ”LLMs suffer from degraded performance when forced to use information within the middle of its long input context”, showing ”lost in the middle” phenomenon where performance degrades significantly when buried in long context.”
Adaptive PII Detection via Evolutionary Orchestration and Multi-agent Debate
Bang, G., Kim, D., Kim, D., & Min, M. (2026) · Lecture notes in computer science · doi:10.1007/978-981-92-1947-6_42
Evaluating LLM-Based Retrieval-Augmented Generation for Soil Science Question Answering
Topić, K., Babac, M. B., & Mornar, V. (2026) · Information · doi:10.3390/info17090859
Zhou, Z., He, Y., Guo, Y., Yu, H., Fang, L., & Li, F. (2026) · International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR) · doi:10.1145/3805712.3809572
A Prototype of AI-Generated Network Protocol and Capability Maturity Evaluation
Wang, Y., Wang, Z., Wang, W., & Wu, W. Q. (2025) · International Conference on Wireless Communications and Signal Processing (WCSP) · doi:10.1109/wcsp68525.2025.1010485
“A critical constraint of current LLMs is limited input/output token capacity [8]: mainstream models (e.g., GPT series) use fixed context windows, so text exceeding this threshold faces (1) irreversible truncation and (2) degraded long-range dependency modeling, harming coherence [9].”
A practical risk framework for large language model use in life science research
Sharpton, T. J., Davis, E. W., & Alexiev, A. (2026) · PLoS Computational Biology · doi:10.1371/journal.pcbi.1014776
“Even within a large window, models attend unevenly across long inputs [11], so fitting material into the window does not guarantee the model uses it.”
Raheem, E. (2026) · SSRN Electronic Journal · doi:10.2139/ssrn.6976218
Tang, Y., Liu, Z., Chen, Z., Yang, J., Xiang, L., Xu, Z., & He, Z. (2026) · Research Square · doi:10.21203/rs.3.rs-9875246/v1
Jarsania, J., & Patel, V. (2025) · International Conference on Artificial Intelligence, Blockchain, Cloud Computing, and Data Analytics (ICoABCD) · doi:10.1109/icoabcd67551.2025.11470747
“Recent work by Liu et al. [3] demonstrates that hallucination rates correlate strongly with context position, peaking at 45% for middle-positioned facts in 128K token contexts.”
Li, H., Ting, H.-K., Jhuo, Y.-C., Chen, C.-L., Kao, C., Yang, M.-H., Tsao, C., Meng, E., Wu, S.-T., & Chiang, P.-J. (2026) · World Journal of Urology · doi:10.1007/s00345-026-06453-7
Kang, J., Zhang, X., Zhai, Y., Feng, D., Ding, B., & Wang, H. (2026) · IEEE International Conference on Joint Cloud Computing (JCC) · doi:10.1109/jcc72984.2026.00052
“Context engineering studies show that long or noisy contexts can degrade LLM performance through irrelevant information and attention dilution [13], [22], [23].”
Enhancing Annotated Bibliography Generation with LLM EnsemblesPreprint
Bermejo, S. (2025) · Qeios · doi:10.32388/wihg9q · Cited by 3
Stochastic Retrieval for Fairness–Accuracy Trade-offs in Clinical RAG
Lunardi, R., Mea, V. D., Eickhoff, C., & Roitero, K. (2026) · ACM International Conference on Bioinformatics, Computational Biology and Biomedicine · doi:10.1145/3807503.3819374
A Retrospective and Adaptive Framework to Improve LLMs
Chang, E. Y. (2025) · ACM eBooks · doi:10.1145/3749421.3749439
“Large windows can inflate the conditioning budget k , and models can still struggle to use mid-context evidence [Liu et al. 2024].”
Chen, H.-C., Ye, C.-F., Hsu, P.-Y., et al. (2026) · Lecture notes on data engineering and communications technologies · doi:10.1007/978-3-032-23260-1_24
ChatAcadien: A RAG-LLM-Based Chatbot for Exploring Acadian Genealogy
Ghali, R., & Selouani, S. (2025) · IEEE International Conference on Collaborative Advances in Software and Computing (CASCON) · doi:10.1109/cascon66301.2025.00026 · Cited by 1
“This approach ensures that each chunk accurately captures its information and helps prevent the ”lost in the middle” phenomenon [24], where LLMs struggle with coherence and relevance when input lengths approach or exceed their maximum context window.”
NER in the courtroom: A data-driven framework for legal entity extraction
Romano, S. P., Sperlí, G., Varlese, M., & Vignali, A. (2026) · Information Processing & Management · doi:10.1016/j.ipm.2026.105132
Brakes, M. F. C., Ferreira, D. O. C., Silva, J. P. C., Novais, A. M. A., Presa, J. P. C., & de Oliveira, S. S. T. (2025) · Escola Regional de Informática de Goiás (ERI-GO) · doi:10.5753/erigo.2025.17152
“Além disso, efeitos de posicio-namento em janelas longas ( lost in the middle ) reforçam a necessidade de um retriever que priorize, no topo, as evidências mais relevantes [Liu et al. 2024].”
Gensheimer, M. F., Adhikari, R., Parmer, C. M., Liu, N., Ma, S. P., & Shieh, L. Y. (2026) · medRxiv · doi:10.64898/2026.06.17.26355917
Self-citationFrom Single to Multi-agent Reasoning: Advancing GeneGPT for Genomics QA
Abedini, K., Shami, F., & Silvello, G. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-21300-6_49
Label space reduction for transductive zero-shot classification with large language models
Vandemoortele, N., Steenwinckel, B., Ongenae, F., & Hoecke, S. V. (2026) · Discover Computing · doi:10.1007/s10791-026-10420-6
“This shared attention makes it harder to focus strongly on relevant information, as demonstrated in “Needle in the haystack" tests [8, 9] where LLMs struggle to retrieve key facts from long contexts [9, 10].”
Zhi, Y., & Ling, W. (2026) · International Journal of Pattern Recognition and Artificial Intelligence · doi:10.1142/s0218001426530022
Dynamic prompt scheduling and context co-optimization for domain-oriented knowledge extraction
Cong, Z., Jin, X., Huang, X., Li, T., Li, Q., Lin, J., Wang, L., Song, D., Liu, Y., Yuan, Y., et al. (2026) · JOURNAL OF SHENZHEN UNIVERSITY SCIENCE AND ENGINEERING · doi:10.3724/sp.j.1249.2026.05533
Design-First Governance for Reliable AI-Assisted Software Development
Meawad, F. E. (2026) · IEEE International Conference on Software Architecture Companion (ICSA-C) · doi:10.1109/icsa-c68850.2026.00048
“Liu et al. demonstrate that even 128k-token windows degrade to 20% recall when critical information is buried [10].”
InfluentialStructured Extraction vs Raw Chunks: Retrieval Strategies for Legal Cases
Pedrozo, D. M., de Oliveira, V. T., Leite, G. D. A., et al. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-39898-7_27
Purifying Multimodal Retrieval: Fragment-Level Evidence Selection for RAG
Wang, X., Wang, Z., Huang, C., Liu, C., Zeng, K., Sheng, Q. Z., & Yao, L. (2026) · International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR) · doi:10.1145/3805712.3809692 · Cited by 1
Wijesiriwardene, T., Thirunarayanan, K., & Sheth, A. (2025) · IEEE Intelligent Systems · doi:10.1109/mis.2025.3621940
“However, current LLMs exhibit a well-documented positional bias , 5 performing best when relevant information appears at the beginning or end of a long context, while often neglecting content located in the middle.”
End-To-End ML with LLMs and Semantic Data Management: Experiences from Chemistry 4.0
Hoseini, S., Herrmann, V., & Quix, C. (2025) · Workshop on Data Management for End-to-End Machine Learning (DEEM) · doi:10.1145/3735654.3735942
“As Liu et al. [33] show, larger models exhibit superior capability of parsing lengthy sequences as compared to small models.”
Wiranata, J. A., & Girsang, A. S. (2026) · International Conference on Smart Science and Technology (IncoSST) · doi:10.1109/incosst69426.2026.11709604
“Relies on labeled question–SQL examples, increasing prompt length and context-window pressure [11], [12].”
Adaptive Context Compression for Retrieval-Augmented Generation
More, R. (2026) · International Conference on Cyber Security, Artificial Intelligence and Digital Economy (CSAIDE) · doi:10.1145/3819836.3819877
“Hierarchical encoding schemes create multi-resolution representations, allowing models to access both high-level summaries and detailed content as needed [5].”
The impact of Software Model Slicing on Software Model Completion with Large Language Models
Welter, A., Bliem, B., Iqbal, O., & Apel, S. (2026) · ACM/IEEE International Conference on Model Driven Engineering Languages and Systems (MODELS) · doi:10.1145/3822455.3830319
“These findings align with prior work suggesting that large inputs can hinder the effective use of task-relevant information even when the relevant information is present in the context [29, 44, 51].”
Scalable Multi-vector Retrieval for Policy Enforcement on E-commerce Marketplaces
Sarpal, A., Kotamraju, S., & Uppalapati, R. (2026) · International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR) · doi:10.1145/3805712.3808506
“Research indicates that performance degrades significantly when models are forced to retrieve information from the middle of long contexts, a phenomenon known as the "lost-in-the-middle" effect [8].”
Task Cascades for Efficient Unstructured Data Processing
Shankar, S., Zeighami, S., & Parameswaran, A. (2026) · Proceedings of the ACM on Management of Data · doi:10.1145/3786702 · Cited by 13
“Per task in the cascade, we need to set confidence thresholds that determine whether to accept the proxy model output for that task.”
ORCASK: A Multi-Agent Tool for Automated Network Configuration of Open Network Switches
HOSSAINI, M. H., Kassler, A., & Bhatt, K. (2026) · IEEE/IFIP Network Operations and Management Symposium (NOMS) · doi:10.1109/noms69089.2026.11668126
“This behavior indicates a positional bias in the baseline approach, where functions located near the beginning or end of the context are more likely to be selected [8].”
RS-Claw: Progressive Active Tool Exploration via Hierarchical Skill Trees for Remote Sensing Agents
Liu, L., Wang, Z., Li, Z., Ouyang, K., Tang, Z., Liu, C., Li, H., Yu, H., Yang, W., Yang, C., et al. (2026) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2026.3720894 · Cited by 5
“Consequently, the model su ff ers from “attention defocus” and the “lost in the middle” phenomenon, rendering it highly susceptible to “tool hallucination” when parsing user intents [15], [16].”
Enhancing static vulnerability alert validation using large language models
Chen, S.-S., Hsu, Y.-S., Lin, T.-C., et al. (2026) · Journal of Systems and Software · doi:10.1016/j.jss.2026.113040
Zhang, T., Mao, X., Zhao, Y., Yang, K., Zhang, Z., Lu, Y., Chen, Y., & Yu, Y. (2026) · ACM Transactions on Software Engineering and Methodology · doi:10.1145/3849809
“Meanwhile, their limited context window and distractible nature [36, 47, 59] also prevent LLMs from learning all repository-level knowledge in the inference phase.”
Risk-Aware Medical Question Answering as a Reliability-Critical Decision Process
Luo, Y., & Li, Y. (2026) · International Conference on Multimedia Retrieval · doi:10.1145/3805622.3810784
“from uncontrolled retrieval ranking under evidence failure[23]. propose CRR, which adds counterfactual constraints on the ranking distribution to stabilize confidence behavior when evidence is Let the passage-side projection be z = 𝑓 𝑝 ( 𝑝 ) .”
Lost in the Evidence? Reproducing Document Position and Context Size Effects in RAG
Gabín, J., Pérez, A., & Parapar, J. (2026) · International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR) · doi:10.1145/3805712.3808569 · Cited by 2
“The lost in the middle phenomenon shows that moving the same answer-bearing content to different positions can change QA accuracy, often peaking near the beginning or end and degrading in the middle [12].”
InfluentialKhemmarach, S., Aewsrisakul, K., Somboon, N., et al. (2025) · IEEE International Conference on Teaching, Assessment, and Learning for Engineering · doi:10.1109/tale66047.2025.11346459
“This aligns with findings that LLMs under-attend to middle-context tokens, even in long-input settings [25].”
Dual-perspective cognitive architecture for long-term dialogue memory
Wang, G., Nie, C., Wang, Z., Zhai, Y., Zhou, J., Haitao, S., & Zhang, H. (2026) · Neurocomputing · doi:10.1016/j.neucom.2026.134419 · Cited by 1
МЕТОДОЛОГІЯ CONTEXT ENGINEERING ДЛЯ ПІДВИЩЕННЯ РЕЛЕВАНТНОСТІ ШІ АГЕНТІВ У SaaS СИСТЕМАХ
Романків, Н. Д. (2025) · ГРААЛЬ НАУКИ · doi:10.36074/grail-of-science.12.12.2025.092 · Cited by 1
“Проте, пряме використання LLM стикається з проблемою "обмеженої пам'яті" (Context Window limit) та "розмиття уваги" (Lost -in-the-middle phenomenon) [1].”
LLM-Supported Hybrid Reranking Architecture for Candidate-Job Matching in Recruitment Processes
Dursun, M., Doğan, B., & Canlı, H. (2026) · Celal Bayar Üniversitesi Fen Bilimleri Dergisi · doi:10.18466/cbayarfbe.1861278
Agent Evolution: Capability-Mechanism Co-evolution
Chen, Y., Zhang, C., Zheng, X., & Tang, S. (2026) · International Conference on Generative Artificial Intelligence and Information Security (GAIIS) · doi:10.1109/gaiis69281.2026.11519194
“As tool library size grows, lengthy text descriptions increase token consumption and induce the model’s “attention diffusion” effect—Liu et al.’s Lost in the Middle phenomenon shows that model response accuracy for tool descriptions in the middle of context drops sharply [9].”
Context Engineering for AI ‐Assisted Pharmacometrics: A Practical Tutorial
Pritchard‐Bell, A., Lin, C., Holmes, W. R., & Doshi, S. (2026) · CPT Pharmacometrics & Systems Pharmacology · doi:10.1002/psp4.70317
“Pharmacometric analysis is central to drug development but requires specialized quantitative training and substantial analyst time [1–4].”
Mapping text to multiplex graph: Prompt compression as Lévy walk-guided graph pruning
Gao, Y., Lu, Y., Deng, J., Nie, J., Tang, Z., Zhang, J., Zhu, Z., Yu, S., Xuan, Q., & Zhou, J. T. (2026) · Knowledge-Based Systems · doi:10.1016/j.knosys.2026.116661
Алпатов, А. Н. (2026) · Современные наукоемкие технологии (Modern High Technologies) · doi:10.17513/snt.40812
Chen, W., Guo, Z., Wang, K., Chang, X., Fan, Q., Chen, J., & Wu, F. (2026) · Expert Systems with Applications · doi:10.1016/j.eswa.2026.133998
TimeX: Phenotype Onset Extraction from Clinical Narratives
Chen, F., Jiang, S., Nguyen, Q. M., Ta, C. N., Wang, K., Chung, W. K., & Weng, C. (2026) · npj Health Systems · doi:10.1038/s44401-026-00099-8
“Thistruncationstrategywasdesigned to mitigate the potential lost-in-the-middle problem 48 , where LLMs tend to .”
Talebian, S. H., Ibrahim, A. O., & Syed, H. J. (2025) · International Conference on Advances in Machine Intelligence and Cybersecurity Technologies (AMICT) · doi:10.1109/amict65811.2025.11402818 · Cited by 1
“Context Window Limitations persist as while recent models extend to >1M tokens, maintaining coherence and avoiding middle information loss remains challenging [40].”
Hallucinations in LLMs: a lifecycle-based survey of causes, detection, mitigation, and prevention
Lamba, N., Tiwari, S., & Gaur, M. (2026) · International Journal of Data Science and Analytics · doi:10.1007/s41060-026-01214-6 · Cited by 1
“• Knowledge Grounding: Knowledge grounding makes sure the generated text is consistent with established bodies of knowledge or other sources of knowledge [58].”
Graph-Aware Reinforcement Learning for Reusable Prompt Compression in Black-Box LLMsPreprint
Eskandarpour, M., Haddadnejad, P., & Jannati, M. J. (2026) · Research Square · doi:10.21203/rs.3.rs-10458568/v1
RAGExplorer: A Visual Analytics System for the Comparative Diagnosis of RAG Systems
Tian, H., Feng, Y., Wen, Z., Li, H., Zhu, M., & Chen, W. (2026) · IEEE Transactions on Visualization and Computer Graphics · doi:10.1109/tvcg.2026.3694443 · Cited by 5
“Theoretical work has identified seven common failure points [7] and highlighted LLM limitations, such as performance degradation with more documents [34] and the “Lost in the Middle” effect [38].”
Kalyanaraman, S., Gupta, R., Joshi, A. S., Joshi, K. S., Suryawanshi, J., & Jadhav, V. (2026) · Systems and Information Engineering Design Symposium · doi:10.1109/sieds69358.2026.11540259
“We hypothesize that larger context windows induce the “lost in the middle” phenomenon [28], where models attend preferentially to context boundaries and generate beyond the evidence.”
Qu, Q., Gao, J., Guo, F., Duffy, V. G., & Man, S. S. (2026) · International Journal of Industrial Ergonomics · doi:10.1016/j.ergon.2026.103996
Bottom-up competition based on label hierarchy trees for in-context learning with large label space
Wei, M., Song, R., & Mao, R. (2026) · Information Fusion · doi:10.1016/j.inffus.2026.104727
The Perils of Naive Truncation: A Context Ablation Study for Dialogue Summarization on DialogSum
Rafi, S., Boddu, L. V. S. R. K., Kolla, D. V., Pothuguntla, C., & Myla, R. B. (2025) · International Conference on Artificial Intelligence and Signal Processing (AISP) · doi:10.1109/aisp68263.2025.11396283
“• BART-large is more robust yet still loses ROUGE-L and factual accuracy, consistent with “lost-in-the-middle” bias [14].”
MAsFL: Data-Secure, Efficient and Accurate Fault Localization with Multi-Agent Small Language Models
Duc, D. P., Sato, H., & Kubo, M. (2026) · International Workshop on Large Language Models for Code (LLM4Code) · doi:10.1145/3786181.3788717
Large Language Models for Binary Decompilation: Paradigms, Evaluation, and Challenges
Wang, Y., Tian, Y., Wei, H., & Wang, F. (2026) · International Conference on Computer Network Security and Software Engineering (CNSSE) · doi:10.1145/3803633.3803666 · Cited by 2
“However, context window limitations [23] make it difficult for LLMs to ingest the information of an entire binary file at once.”
Zhao, Y., He, Y., Zhang, X., et al. (2026) · Neurocomputing · doi:10.1016/j.neucom.2025.132577
Кузнецов, И. (2026) · Международный научно-исследовательский журнал · doi:10.60797/irj.2026.171.79
Predicting the Next Words From Shannon's Entropy to Large Language ModelsPreprint
Osterrieder, J. R. (2026) · SSRN Electronic Journal · doi:10.2139/ssrn.6616078
Qian, B., Wu, Y., Zeng, S., Wang, Z., & 王乔晨 (2026) · Lecture notes in computer science · doi:10.1007/978-981-92-1300-9_45
Reversing the Death of Hypertext: Mechanistic Interpretability for LLM Navigation
Zheng, W., He, Y., & Li, J. (2026) · ACM SIGWEB Newsletter · doi:10.1145/3801080.3801084
A Structural Survey of Coordination Patterns in LLM-Based Multi-agent Systems
Esposito, N., Aniello, G., Gambino, G., Panariello, C., Alfano, D., & Abbruzzese, R. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-40774-0_21
Zhang, Y., Li, M., Lu, Z., Cai, Z., & Shan, J. (2026) · Sensors · doi:10.3390/s26123844
“Therefore, the formulation itself is only one part of the task; the surrounding system must also determine whether the produced model and result are trustworthy [3,13].”
Toward trustworthy engineering information extraction using retrieval-augmented generation
Shen, J., Zhu, H., & Zhang, B. (2026) · Knowledge-Based Systems · doi:10.1016/j.knosys.2026.116418 · Cited by 1
Lu, G., & Wu, T. (2026) · International Conference on Machine Learning, Natural Language Processing and Modeling · doi:10.1145/3832928.3833023
Using Knowledge Graph To Detect And Explain Misinformation Spread On The Web
Peskine, Y. (2025) · theses.fr (ABES) · doi:10.70675/efad38dcza422z419dz865ez1bc0e700ce2c
SSFGraphRAG: Semantic-Structure Fusion Graph Retrieval-Augmented Generation
Liu, B., Wang, M.-H., Wang, J., Xu, B., Gan, J., & Han, X. (2025) · International Conference on Vision, Image and Signal Processing (ICVISP) · doi:10.1109/icvisp68610.2025.11451238
“However, KG-RAG faces challenges: retrieving multi-hop subgraphs and balancing efficiency with coverage, causing the "lost-in-the-middle" phenomenon [10].”
Concept-based heterogeneous context fusion for enhancing RAG via AMR-driven distillation
Shi, K., Sun, X., Lin, Q., Li, Y., Li, Q., & Xu, G. (2026) · Pattern Recognition · doi:10.1016/j.patcog.2026.114604
Kim, R. S. (2026) · Discover Artificial Intelligence · doi:10.1007/s44163-026-01573-w
“Analogous dynamics have been reported in transformer-based systems: attention-pattern dispersion, self-attention limits, and effective-context shrinkage under long or adversarial inputs can reduce a model’s capacity to maintain global narrative dependencies [20, 46, 47].”
InfluentialDeep reasoning of large models based on knowledge graph
Cao, C., Du, Z., & Li, X. (2025) · International Conference on Graphics and Image Processing (ICGIP) · doi:10.1117/12.3057811
“However, in practical scenarios, RAG often encounters the "Lost in the Middle" [4] dilemma, where the repeated content in the form of text fragments during concatenation into prompts results in excessively long contexts.”
A Neutral and Evidence-Grounded Evaluation of Chunking Strategies in Industrial RAG Systems
Aguiar, V. A., Pereira, G. L. B., Novais, A. M. A., de Jesus, T. P., de Oliveira, S. S. T., & Soares, A. S. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-39904-5_13
Token Cost Optimization in an LLM Agent for JSON-Based System Modeling from Dialogue History
Lavrentiev, S., & Dukhanov, A. V. (2026) · International Russian Smart Industry Conference (SmartIndustryCon) · doi:10.1109/smartindustrycon68821.2026.11493004 · Cited by 1
“This “sliding window” is easy to implement and effectively limits request length, but it leads to the loss of important information, especially in stepwise modeling of complex systems where early assumptions and definitions remain relevant throughout the entire session [9].”
InfluentialBeyond Rows to Reasoning: Agentic Retrieval for Multimodal Spreadsheet Understanding
Gulati, A., Sen, S., Sarguroh, W., & Paul, K. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-36023-6_33
Zhao, X., Feng, Y., Sun, D., Gao, L., Xu, T., Lv, J., Zhou, G., & Xia, C. (2026) · International Conference on Machine Intelligence and Digital Applications (MIDA) · doi:10.1145/3801438.3803160
“On the cost front, HELP’s vector-based log grouping is making LLM-powered parsing increasingly realistic for production-scale deployment [46].”
Non-invasive Generative AI in Moodle via RPA and Retrieval-Augmented Generation
Rohanadeera, S. L., & Hewagamage, K. P. (2026) · Communications in computer and information science · doi:10.1007/978-981-95-9843-4_1
Park, C. M., Park, I. H., Kim, M., Jang, Y., Lim, H., & Hur, Y. (2026) · JCO Clinical Cancer Informatics · doi:10.1200/cci-25-00361 · Cited by 1
“…beyond optimal retrieval levels, with rubric scores declining 64% from 3.77 at k 5 5 to 1.37 at k 5 50, attributable to the “ lost-in-the-middle ” phenomenon 49 and the concentrated nature of clinical queries — relevant information resides in few chunks while additional passages dilute precision.”
InfluentialKenlie, E. A., & Anderies (2026) · Procedia Computer Science · doi:10.1016/j.procs.2026.08.329
From the First Public ChatGPT to Astra in Scientific Collaboration
Shapovalov, V. (2026) · Conceptual Future · doi:10.53933/c75rcf79 · Cited by 2
SPARQ: A Cost-Efficient Framework for Offline Table Question Answering via Adaptive Routing
Liu, Y., Yan, M., Xue, J., Ren, W., Ye, Y., Zhou, H., Li, J., & Chen, Z. (2026) · IEEE International Conference on Data Engineering (ICDE) · doi:10.1109/icde65706.2026.00199
“…directly substituting offline LLMs into pipelines designed for online models leads to critical failure modes: (1) Even though offline LLMs may support large context windows ( i.e., the input can fit), their effective reasoning range over long or complex tables is often much smaller [45], [63], [64].”
Tanhaei, M. (2026) · Results in Engineering · doi:10.1016/j.rineng.2026.113085 · Cited by 1
Entropy-Guided Telemetry Compression for LLM-Based RCA: An Industrial Pilot on Log Data
García-Hernández, C. F., Abdali, A., Wu, G., Pang, Z., Shen, F., Zhang, Y., & Wang, M. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37685-5_51
Agent-Driven Topology Assembly: Trustworthy LLM Dispatch Reasoning in Distribution Networks
Yang, H., Li, Z., Yin, R., et al. (2026) · Asia Conference on Power and Electrical Engineering (ACPEE) · doi:10.1109/acpee69242.2026.11525045
“Furthermore, multi-hop topological tracing rapidly bloats the LLM's context window, diluting attention [10] and degrading the Reasoner's final synthesis.”
REM-CaR: Retrieval-Augmented Generation Enhanced Medical Case-Based Reasoner
Li, Z., Jin, B., Chen, Y., Gong, T., & Fu, Y. (2026) · Lecture notes in computer science · doi:10.1007/978-981-92-5699-0_35
PRADA-QA: Product QA with Multi-Agent Planning and Dynamic Knowledge Retrieval
Nguyen, H.-D., Do, T.-D., Nguyễn, V. C., Tran, O. T., & Le, D.-T. (2026) · Communications in computer and information science · doi:10.1007/978-981-92-2590-3_24
Towards Automated Parameter Extraction from Engineering Documents
Möller, C., Björkskog, C., Jatta, L., Lundell, A., Manngård, M., & Westö, J. (2025) · IFAC-PapersOnLine · doi:10.1016/j.ifacol.2025.11.864
Wilson, C., & Purwanto, E. S. (2026) · International Conference on Information Management and Technology (ICIMTech) · doi:10.1109/icimtech202671078.2026.11698989
“RankGPT [15] showed that zero-shot LLM listwise ranking matches supervised cross-encoders, and permutation self-consistency [11] mitigates the position bias [12] of such listwise prompts.”
NanoKnow: How to Know What Your Language Model Knows
Gu, L., Jedidi, N., & Lin, J. (2026) · International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR) · doi:10.1145/3805712.3808604 · Cited by 2
“However, the rate of this improvement, especially for SQuAD, is much lower than in the closed-book setting, demonstrating that RAG can help mitigate this dependence on pre-training frequency.”
Guan, B., Cui, W., & Juhász, L. (2026) · Transactions in GIS · doi:10.1111/tgis.70402 · Cited by 4
LLM-Enabled Multi-Agent Collaborative Instrument Control for Automated Microelectronic Testing
Zhang, Y., Liu, J., Zhang, Z., qi, L., Wang, P., Wang, X., & Zhao, W. S. (2026) · Advanced Devices & Instrumentation · doi:10.34133/adi.0202 · Cited by 2
Formalizing Requirements into Dafny Specifications with LLMs
Lu, Y.-H., Zhu, X.-Y., Zhang, W., & Yan, R. (2025) · Lecture notes in computer science · doi:10.1007/978-981-95-4213-0_6 · Cited by 2
Rao, A., Bali, A., Tran, V. V., Vasavada, S., Chu, J., Kanwal, F., Rusin, C., & El‐Serag, H. B. (2026) · Clinical Gastroenterology and Hepatology · doi:10.1016/j.cgh.2026.05.017 · Cited by 1
Metadata
- Authors
- Liu, N. F., Lin, K., Hewitt, J., Paranjape, A., Bevilacqua, M., Petroni, F., & Liang, P.
- Affiliations
- Stanford University (US)
- University of California, Berkeley (US)
- Samaya AI, UK
- Corresponding author
- Liu, Nelson F.
- Venue
- Transactions of the Association for Computational Linguistics
- Publisher
- MIT Press
- Published
- 2024
- Volume
- 12
- Pages
- 157-173
- ISSN
- 2307-387X (ISSN-L)
- OpenAlex
- W4391876619
- Semantic Scholar
- 1733eb7792f7a43dd21f51f4d1017a1bffd217b5
- Cited by
- 1,612
- References
- 59
- Citations by year
- 2026: 901 · 2025: 516 · 2024: 172
- FWCI
- 233.95
- Type
- journal article
- Language
- en
- Open access
- diamond
- doi.org
- Licence
- Locations
- Transactions of the Association for Computational Linguistics (publishedVersion, cc-by)
- arXiv (Cornell University) (submittedVersion)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- arXiv (Cornell University)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Topic Modeling
- Physical Sciences › Computer Science › Artificial Intelligence › Natural Language Processing Techniques
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Multimodal Machine Learning Applications
- Keywords
- long-context language models, multi-document question answering, position bias
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-09-27 (prompt extract-v2).
Cited sources are checked with a language model’s help (gpt-6-sol).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.