{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,4,14]],"date-time":"2026-04-14T17:28:53Z","timestamp":1776187733278,"version":"3.50.1"},"reference-count":50,"publisher":"Elsevier BV","license":[{"start":{"date-parts":[[2026,6,1]],"date-time":"2026-06-01T00:00:00Z","timestamp":1780272000000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/www.elsevier.com\/tdm\/userlicense\/1.0\/"},{"start":{"date-parts":[[2026,6,1]],"date-time":"2026-06-01T00:00:00Z","timestamp":1780272000000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/www.elsevier.com\/legal\/tdmrep-license"},{"start":{"date-parts":[[2026,6,1]],"date-time":"2026-06-01T00:00:00Z","timestamp":1780272000000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-017"},{"start":{"date-parts":[[2026,6,1]],"date-time":"2026-06-01T00:00:00Z","timestamp":1780272000000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"},{"start":{"date-parts":[[2026,6,1]],"date-time":"2026-06-01T00:00:00Z","timestamp":1780272000000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-012"},{"start":{"date-parts":[[2026,6,1]],"date-time":"2026-06-01T00:00:00Z","timestamp":1780272000000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2026,6,1]],"date-time":"2026-06-01T00:00:00Z","timestamp":1780272000000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-004"}],"funder":[{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["62306216"],"award-info":[{"award-number":["62306216"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100012226","name":"Fundamental Research Funds for the Central Universities","doi-asserted-by":"publisher","award":["2042025kf0026"],"award-info":[{"award-number":["2042025kf0026"]}],"id":[{"id":"10.13039\/501100012226","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":["elsevier.com","sciencedirect.com"],"crossmark-restriction":true},"short-container-title":["Neural Networks"],"published-print":{"date-parts":[[2026,6]]},"DOI":"10.1016\/j.neunet.2026.108720","type":"journal-article","created":{"date-parts":[[2026,2,13]],"date-time":"2026-02-13T00:42:39Z","timestamp":1770943359000},"page":"108720","update-policy":"https:\/\/doi.org\/10.1016\/elsevier_cm_policy","source":"Crossref","is-referenced-by-count":0,"special_numbering":"C","title":["Generalized and group spherical linear interpolation for token-level context compression"],"prefix":"10.1016","volume":"198","author":[{"ORCID":"https:\/\/orcid.org\/0009-0005-7894-7549","authenticated-orcid":false,"given":"Jinhao","family":"Tian","sequence":"first","affiliation":[]},{"ORCID":"https:\/\/orcid.org\/0000-0003-0436-8446","authenticated-orcid":false,"given":"Zuchao","family":"Li","sequence":"additional","affiliation":[]},{"ORCID":"https:\/\/orcid.org\/0000-0001-8275-3070","authenticated-orcid":false,"given":"Meng-Jia","family":"Shen","sequence":"additional","affiliation":[]},{"ORCID":"https:\/\/orcid.org\/0000-0003-0542-2280","authenticated-orcid":false,"given":"Lefei","family":"Zhang","sequence":"additional","affiliation":[]}],"member":"78","reference":[{"key":"10.1016\/j.neunet.2026.108720_bib0001","unstructured":"Bahdanau, D., Cho, K., & Bengio, Y. (2014). Neural machine translation by jointly learning to align and translate. CoRRarXiv: 1409.0473."},{"key":"10.1016\/j.neunet.2026.108720_bib0002","unstructured":"Bai, Y., Lv, X., Zhang, J., Lyu, H., Tang, J., Huang, Z., Du, Z., Liu, X., Zeng, A., Hou, L., & et al. (2023). Longbench: A bilingual, multitask benchmark for long context understanding. arXiv: 2308.14508."},{"key":"10.1016\/j.neunet.2026.108720_bib0003","unstructured":"Bolya, D., Fu, C.-Y., Dai, X., Zhang, P., Feichtenhofer, C., & Hoffman, J. (2022). Token merging: Your vit but faster. arXiv: 2210.09461."},{"key":"10.1016\/j.neunet.2026.108720_bib0004","series-title":"Proceedings of the 2022 conference on empirical methods in natural language processing","first-page":"10114","article-title":"A dataset for hyper-relational extraction and a cube-filling approach","author":"Chia","year":"2022"},{"key":"10.1016\/j.neunet.2026.108720_bib0005","series-title":"Proceedings of the third international workshop on paraphrasing (iwp2005)","article-title":"Automatically constructing a corpus of sentential paraphrases","author":"Dolan","year":"2005"},{"key":"10.1016\/j.neunet.2026.108720_bib0006","series-title":"Gpt-3: Its nature, scope, limits, and consequences","first-page":"681","volume":"vol. 30","author":"Floridi","year":"2020"},{"key":"10.1016\/j.neunet.2026.108720_bib0007","unstructured":"Ge, S., Zhang, Y., Liu, L., Zhang, M., Han, J., & Gao, J. (2023). Model tells you what to discard: Adaptive kv cache compression for llms. arXiv: 2310.01801."},{"key":"10.1016\/j.neunet.2026.108720_bib0008","first-page":"3991","article-title":"Lm-infinite: Zero-shot extreme length generalization for large language models","author":"Han","year":"2024"},{"key":"10.1016\/j.neunet.2026.108720_bib0009","series-title":"International conference on learning representations","article-title":"Deberta: Decoding-enhanced bert with disentangled attention","author":"He","year":"2021"},{"key":"10.1016\/j.neunet.2026.108720_bib0010","series-title":"Proceedings of the 60th annual meeting of the association for computational linguistics (Volume 1: Long papers)","first-page":"3774","article-title":"Token dropping for efficient BERT pretraining","author":"Hou","year":"2022"},{"key":"10.1016\/j.neunet.2026.108720_bib0011","series-title":"Proceedings of the 2022 conference on empirical methods in natural language processing","first-page":"8826","article-title":"Promptbert: Improving bert sentence embeddings with prompts","author":"Jiang","year":"2022"},{"key":"10.1016\/j.neunet.2026.108720_bib0012","doi-asserted-by":"crossref","first-page":"73","DOI":"10.1016\/j.jbi.2015.03.010","article-title":"Cadec: A corpus of adverse drug event annotations","volume":"55","author":"Karimi","year":"2015","journal-title":"Journal of Biomedical Informatics"},{"key":"10.1016\/j.neunet.2026.108720_bib0013","series-title":"International conference on learning representations","article-title":"Albert: A lite bert for self-supervised learning of language representations","author":"Lan","year":"2020"},{"key":"10.1016\/j.neunet.2026.108720_bib0014","unstructured":"Liu, Y., Ott, M., Goyal, N., Du, J., Joshi, M., Chen, D., Levy, O., Lewis, M., Zettlemoyer, L., & Stoyanov, V., (2019). Roberta: A robustly optimized bert pretraining approach. arXiv: 1907.11692."},{"key":"10.1016\/j.neunet.2026.108720_bib0015","article-title":"Scissorhands: Exploiting the persistence of importance hypothesis for llm kv cache compression at test time","volume":"36","author":"Liu","year":"2024","journal-title":"Advances in Neural Information Processing Systems"},{"key":"10.1016\/j.neunet.2026.108720_bib0016","series-title":"Proceedings of the 60th annual meeting of the association for computational linguistics (Volume 1: Long papers)","article-title":"Unified structure generation for universal information extraction","author":"Lu","year":"2022"},{"key":"10.1016\/j.neunet.2026.108720_bib0017","series-title":"Proceedings of the 2018 conference on empirical methods in natural language processing","first-page":"3219","article-title":"Multi-task identification of entities, relations, and coreference for scientific knowledge graph construction","author":"Luan","year":"2018"},{"key":"10.1016\/j.neunet.2026.108720_bib0018","unstructured":"Luohe, S., Hongyi, Z., Yao, Y., Zuchao, L., & Hai, Z. (2024). Keep the cost down: A review on methods to optimize llm\u2019s kv-cache consumption. arXiv: 2407.18003."},{"key":"10.1016\/j.neunet.2026.108720_bib0019","series-title":"Ace 2004 multilingual training corpus, Web Download, philadelphia: Linguistic Data Consortium","author":"Mitchell","year":"2005"},{"key":"10.1016\/j.neunet.2026.108720_bib0020","first-page":"19327","article-title":"Learning to compress prompts with gist tokens","volume":"36","author":"Mu","year":"2024","journal-title":"Advances in Neural Information Processing Systems"},{"key":"10.1016\/j.neunet.2026.108720_bib0021","doi-asserted-by":"crossref","unstructured":"Pang, J., Ye, F., Wong, D. F., & Wang, L. (2024). Anchor-based large language models. arXiv: 2402.07616.","DOI":"10.18653\/v1\/2024.findings-acl.295"},{"key":"10.1016\/j.neunet.2026.108720_bib0022","series-title":"9th international conference on learning representations, ICLR 2021","article-title":"Structured prediction as translation between augmented natural languages","author":"Paolini","year":"2021"},{"key":"10.1016\/j.neunet.2026.108720_bib0023","series-title":"Proceedings of the 61st annual meeting of the association for computational linguistics (Volume 1: Long papers)","first-page":"16318","article-title":"FSUIE: A novel fuzzy span mechanism for universal information extraction","author":"Peng","year":"2023"},{"key":"10.1016\/j.neunet.2026.108720_bib0024","series-title":"Proceedings of the 10th international workshop on semantic evaluation (semeval-2016)","first-page":"19","article-title":"SemEval-2016 task 5: Aspect based sentiment analysis","author":"Pontiki","year":"2016"},{"key":"10.1016\/j.neunet.2026.108720_bib0025","series-title":"Proceedings of the 9th international workshop on semantic evaluation","first-page":"486","article-title":"Semeval-2015 task 12: Aspect based sentiment analysis","author":"Pontiki","year":"2015"},{"key":"10.1016\/j.neunet.2026.108720_bib0026","series-title":"Proceedings of the 8th international workshop on semantic evaluation","first-page":"27","article-title":"Semeval-2014 task 4: Aspect based sentiment analysis","author":"Pontiki","year":"2014"},{"issue":"140","key":"10.1016\/j.neunet.2026.108720_bib0027","first-page":"1","article-title":"Exploring the limits of transfer learning with a unified text-to-text transformer","volume":"21","author":"Raffel","year":"2020","journal-title":"Journal of Machine Learning Research"},{"key":"10.1016\/j.neunet.2026.108720_bib0028","series-title":"Ecml\/pkdd","first-page":"2386383","article-title":"Modeling relations and their mentions without labeled text","author":"Riedel","year":"2010"},{"key":"10.1016\/j.neunet.2026.108720_bib0029","series-title":"Proceedings of the eighth conference on computational natural language learning","first-page":"1","article-title":"A linear programming formulation for global inference in natural language tasks","author":"Roth","year":"2004"},{"key":"10.1016\/j.neunet.2026.108720_bib0030","series-title":"Proceedings of the seventh conference on natural language learning at hlt-naacl 2003","first-page":"142","article-title":"Introduction to the conll-2003 shared task: language-independent named entity recognition","author":"Sang","year":"2003"},{"key":"10.1016\/j.neunet.2026.108720_bib0031","doi-asserted-by":"crossref","first-page":"8749","DOI":"10.1609\/aaai.v34i05.6401","article-title":"Casie: Extracting cybersecurity event information from text","volume":"34","author":"Satyapanich","year":"2020","journal-title":"Proceedings of the AAAI Conference on Artificial Intelligence"},{"key":"10.1016\/j.neunet.2026.108720_bib0032","series-title":"Proceedings of the 2013 conference on empirical methods in natural language processing","first-page":"1631","article-title":"Recursive deep models for semantic compositionality over a sentiment treebank","author":"Socher","year":"2013"},{"key":"10.1016\/j.neunet.2026.108720_bib0033","series-title":"Proceedings of the international conference on learning representations (ICLR","article-title":"Hierarchical context merging: Better long context understanding for pre-trained llms","author":"Song","year":"2024"},{"key":"10.1016\/j.neunet.2026.108720_bib0034","doi-asserted-by":"crossref","DOI":"10.1016\/j.neucom.2023.127063","article-title":"Roformer: Enhanced transformer with rotary position embedding","volume":"568","author":"Su","year":"2024","journal-title":"Neurocomputing"},{"key":"10.1016\/j.neunet.2026.108720_bib0035","unstructured":"Touvron, H., Martin, L., Stone, K., Albert, P., Almahairi, A., Babaei, Y., Bashlykov, N., Batra, S., Bhargava, P., & Bhosale, S., et al. (2023). Llama 2: Open foundation and fine-tuned chat models. arXiv: 2307.09288."},{"key":"10.1016\/j.neunet.2026.108720_bib0036","first-page":"5998","article-title":"Attention is all you need","volume":"30","author":"Vaswani","year":"2017","journal-title":"Advances in Neural Information Processing Systems"},{"key":"10.1016\/j.neunet.2026.108720_bib0037","series-title":"Ace 2005 multilingual training corpus, Web Download, philadelphia: Linguistic Data Consortium","author":"Walker","year":"2006"},{"key":"10.1016\/j.neunet.2026.108720_bib0038","unstructured":"Wan, Z., Wu, X., Zhang, Y., Xin, Y., Tao, C., Zhu, Z., Wang, X., Luo, S., Xiong, J., & Zhang, M. (2024). Dynamic discriminative operations for efficient generative inference of large language models. arXiv: 2406.13035."},{"key":"10.1016\/j.neunet.2026.108720_bib0039","series-title":"Proceedings of the 2018 emnlp workshop blackboxnlp: Analyzing and interpreting neural networks for NLP","first-page":"353","article-title":"Glue: A multi-task benchmark and analysis platform for natural language understanding","author":"Wang","year":"2018"},{"key":"10.1016\/j.neunet.2026.108720_bib0040","series-title":"Findings of the association for computational linguistics: Acl 2022","first-page":"803","article-title":"Deepstruct: Pretraining of language models for structure prediction","author":"Wang","year":"2022"},{"key":"10.1016\/j.neunet.2026.108720_bib0041","unstructured":"Wang, X., Zhou, W., Zu, C., Xia, H., Chen, T., Zhang, Y., Zheng, R., Ye, J., Zhang, Q., & Gui, T., et al. (2023). Instructuie: Multi-task instruction tuning for unified information extraction. arXiv: 2304.08085."},{"key":"10.1016\/j.neunet.2026.108720_bib0042","unstructured":"Wang, Z., Jin, B., Yu, Z., & Zhang, M. (2024). Model tells you where to merge: Adaptive kv cache merging for llms on long-context tasks. arXiv: 2407.08454."},{"key":"10.1016\/j.neunet.2026.108720_bib0043","doi-asserted-by":"crossref","first-page":"625","DOI":"10.1162\/tacl_a_00290","article-title":"Neural network acceptability judgments","volume":"7","author":"Warstadt","year":"2019","journal-title":"Transactions of the Association for Computational Linguistics"},{"key":"10.1016\/j.neunet.2026.108720_bib0044","article-title":"A broad-coverage challenge corpus for sentence understanding through inference","volume":"vol. 1","author":"Williams","year":"2018"},{"key":"10.1016\/j.neunet.2026.108720_bib0045","unstructured":"Xiao, G., Tian, Y., Chen, B., Han, S., & Lewis, M. (2023). Efficient streaming language models with attention sinks. arXiv: 2309.17453."},{"key":"10.1016\/j.neunet.2026.108720_bib0046","doi-asserted-by":"crossref","unstructured":"Yang, D., Han, X., Gao, Y., Hu, Y., Zhang, S., & Zhao, H. (2024). Pyramidinfer: Pyramid KV cache compression for high-throughput llm inference. arXiv: 2405.12532.","DOI":"10.18653\/v1\/2024.findings-acl.195"},{"key":"10.1016\/j.neunet.2026.108720_bib0047","series-title":"International conference on machine learning","article-title":"Cam: Cache merging for memory-efficient llms inference","author":"Zhang","year":"2024"},{"key":"10.1016\/j.neunet.2026.108720_bib0048","first-page":"34661","article-title":"H2o: Heavy-hitter oracle for efficient generative inference of large language models","volume":"36","author":"Zhang","year":"2024","journal-title":"Advances in Neural Information Processing Systems"},{"key":"10.1016\/j.neunet.2026.108720_bib0049","doi-asserted-by":"crossref","unstructured":"Zhong, Q., Ding, L., Liu, J., Liu, X., Zhang, M., Du, B., & Tao, D. (2023). Revisiting token dropping strategy in efficient bert pretraining. arXiv: 2305.15273.","DOI":"10.18653\/v1\/2023.acl-long.579"},{"key":"10.1016\/j.neunet.2026.108720_bib0050","series-title":"Proceedings of the 2023 conference on empirical methods in natural language processing","first-page":"8861","article-title":"Mirror: A universal framework for various information extraction tasks","author":"Zhu","year":"2023"}],"container-title":["Neural Networks"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/api.elsevier.com\/content\/article\/PII:S0893608026001826?httpAccept=text\/xml","content-type":"text\/xml","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/api.elsevier.com\/content\/article\/PII:S0893608026001826?httpAccept=text\/plain","content-type":"text\/plain","content-version":"vor","intended-application":"text-mining"}],"deposited":{"date-parts":[[2026,4,14]],"date-time":"2026-04-14T16:36:58Z","timestamp":1776184618000},"score":1,"resource":{"primary":{"URL":"https:\/\/linkinghub.elsevier.com\/retrieve\/pii\/S0893608026001826"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,6]]},"references-count":50,"alternative-id":["S0893608026001826"],"URL":"https:\/\/doi.org\/10.1016\/j.neunet.2026.108720","relation":{},"ISSN":["0893-6080"],"issn-type":[{"value":"0893-6080","type":"print"}],"subject":[],"published":{"date-parts":[[2026,6]]},"assertion":[{"value":"Elsevier","name":"publisher","label":"This article is maintained by"},{"value":"Generalized and group spherical linear interpolation for token-level context compression","name":"articletitle","label":"Article Title"},{"value":"Neural Networks","name":"journaltitle","label":"Journal Title"},{"value":"https:\/\/doi.org\/10.1016\/j.neunet.2026.108720","name":"articlelink","label":"CrossRef DOI link to publisher maintained version"},{"value":"article","name":"content_type","label":"Content Type"},{"value":"\u00a9 2026 Published by Elsevier Ltd.","name":"copyright","label":"Copyright"}],"article-number":"108720"}}