---
title: "Hangzhou — research institutions and their papers"
description: "Hangzhou (China): 8 institutions with papers on Elamur Research — 3 universities, 1 research institute, 4 companies — and 41,869 citations of their works."
canonical: "https://research.elamur.ai/free/en/agglomerations/hangzhou/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/agglomerations/hangzhou/)

[home](https://md.research.elamur.ai/free/en/) / [institutions](https://md.research.elamur.ai/free/en/orgs/) / [agglomerations](https://md.research.elamur.ai/free/en/agglomerations/) / Hangzhou

# Hangzhou

Agglomeration [China](https://md.research.elamur.ai/free/en/countries/china/) · metro population 10,129,997 (functional urban area, GHSL 2015)

**8** institutions with a page**92** works in our library**41,869** citations of their works

## About

### Profile

Hangzhou is a city region in [China](https://md.research.elamur.ai/free/en/countries/china/). On this site it appears as a hub of research in natural language processing, large language models and multimodal artificial intelligence. Eight institutions based in the region have works here: three universities, one research institute and four companies. Its industrial laboratories carry as much weight as its universities, and they are, if anything, the more visible part of its research.

The companies are [Alibaba Group](https://md.research.elamur.ai/free/en/companies/alibaba-group/), [Alibaba DAMO Academy](https://md.research.elamur.ai/free/en/companies/alibaba-damo-academy/), [Ant Group](https://md.research.elamur.ai/free/en/companies/ant-group/) and [DeepSeek](https://md.research.elamur.ai/free/en/companies/deepseek/). The universities are [Zhejiang University](https://md.research.elamur.ai/free/en/universities/zhejiang-university/), [Westlake University](https://md.research.elamur.ai/free/en/universities/westlake-university/) and [Hangzhou Dianzi University](https://md.research.elamur.ai/free/en/universities/hangzhou-dianzi-university/). The research institute is [Zhejiang Lab](https://md.research.elamur.ai/free/en/labs/zhejiang-lab/).

Alibaba Group and Zhejiang University have the largest bodies of work in the region on this site, and both are among its most cited institutions. DeepSeek has only a small set of papers here, yet those papers draw an exceptional share of the region's citations. Most of that attention goes to its work on reasoning in language models.

Taken together, the Hangzhou works in our library sit almost entirely within one broad area: computational language and vision. They cover how machines read text and extract information from it, and how they analyse sentiment and opinion. They also cover how language models are trained, adapted, evaluated and made to reason, and how those models are joined to images, video and audio. The picture below is drawn only from the works and people on this site. It describes the region as this library shows it, not its whole scientific output.

### History

The earliest Hangzhou work in this collection comes from Zhejiang University in 2011. [Opinion Word Expansion and Target Extraction through Double Propagation](https://md.research.elamur.ai/free/en/papers/qiu-2011-opinion-word-expansion-target-extraction/) treats opinion mining as two linked tasks. One is finding the words that carry opinion; the other is finding the things those opinions are about. The method spreads knowledge back and forth between the two. Sentiment and opinion analysis begins here, and it stays one of the region's longest-running lines.

A gap follows until 2018. In that year Alibaba Group and Zhejiang University share [TextBugger: Generating Adversarial Text Against Real-world Applications](https://md.research.elamur.ai/free/en/papers/li-2018-textbugger-generating-adversarial-text-against/). The paper studies how text can be altered so that it still reads naturally to people while misleading automated classifiers. It marks the first joint work between the region's largest company and its largest university among the works here. It also opens a concern with robustness that returns years later in research on prompts.

In 2019, Hangzhou Dianzi University appears with [ActivityNet-QA](https://md.research.elamur.ai/free/en/papers/yu-2019-activitynet-qa-dataset-understanding-complex/), a dataset for answering questions about complex web videos. The paper argues that video question answering had lagged behind image question answering because its benchmarks were weaker and smaller. It is the region's first contribution here to video understanding.

The years 2020 and 2021 bring a dense cluster of work in core language processing. Zhejiang University contributes [A Unified MRC Framework for Named Entity Recognition](https://md.research.elamur.ai/free/en/papers/li-2020-unified-mrc-framework-named-entity/) and [Dice Loss for Data-imbalanced NLP Tasks](https://md.research.elamur.ai/free/en/papers/li-2020-dice-loss-data-imbalanced-nlp/). Alibaba DAMO Academy adds [A Joint Neural Model for Information Extraction with Global Features](https://md.research.elamur.ai/free/en/papers/lin-2020-joint-neural-model-information-extraction/) and work on aspect sentiment triplets and quadruples. Alibaba Group contributes a graph attention model for aspect-based sentiment. Westlake University appears with reasoning and dialogue datasets, among them [LogiQA](https://md.research.elamur.ai/free/en/papers/liu-2020-logiqa-challenge-dataset-machine-reading/), [MuTual](https://md.research.elamur.ai/free/en/papers/cui-2020-mutual-dataset-multi-turn-dialogue/) and, in 2021, [DialogSum](https://md.research.elamur.ai/free/en/papers/chen-2021-dialogsum-real-life-scenario-dialogue/).

The year 2023 marks a sharp turn toward large language models, and every institution in the region takes part. Surveys appear on instruction tuning, on the evaluation of language models, and on joining language models with knowledge graphs. Benchmarks appear for multimodal models and for the robustness of prompts. Studies test sentiment analysis, information extraction, logical reasoning and text-to-SQL with the new models. New systems also appear, including vision-language models, audio-visual video models and a medical chat model.

From 2024 onward, the centre of gravity moves to reasoning and to open model families. DeepSeek publishes on mathematical reasoning, process supervision, vision-language understanding and, in 2025, reasoning trained through reinforcement learning, as well as formal theorem proving. Alibaba Group releases a technical report on a code model. Ant Group's works here, both from 2025, explore memory for language-model agents and diffusion as an alternative way of building language models.

### Research profile

Information extraction and named entity recognition form one of the oldest and most continuous lines. At Zhejiang University, the unified machine reading comprehension framework recasts entity recognition as answering questions about a text. Doing so lets a single approach handle both nested and flat entities. Later, [GPT-NER](https://md.research.elamur.ai/free/en/papers/wang-2025-gpt-ner-named-entity-recognition/) asks why large language models, strong on many tasks through in-context learning, fall short on entity recognition, and it proposes a way to close that gap. Zhejiang University and Westlake University share [Template-Based Named Entity Recognition Using BART](https://md.research.elamur.ai/free/en/papers/cui-2021-template-based-named-entity-recognition/), which deals with few-shot recognition when a target domain has little labelled text. At Alibaba DAMO Academy, the joint neural model called ONEIE treats everything extracted from a sentence as one information graph. In the language-model era, [ChatIE](https://md.research.elamur.ai/free/en/papers/wei-2023-chatie-zero-shot-information-extraction/) turns zero-shot extraction into a two-stage, multi-turn conversation with ChatGPT.

Sentiment analysis and opinion mining run from 2011 to 2023. After the early double-propagation method at Zhejiang University, Alibaba Group's [Relational Graph Attention Network for Aspect-based Sentiment Analysis](https://md.research.elamur.ai/free/en/papers/wang-2020-relational-graph-attention-network-aspect/) argues that syntax shaped around the target aspect can fix errors made by general models. Alibaba DAMO Academy takes the line toward ever finer structure. [Position-Aware Tagging for Aspect Sentiment Triplet Extraction](https://md.research.elamur.ai/free/en/papers/xu-2020-position-aware-tagging-aspect-sentiment/) extracts aspect, opinion and polarity together. [Aspect Sentiment Quad Prediction as Paraphrase Generation](https://md.research.elamur.ai/free/en/papers/zhang-2021-aspect-sentiment-quad-prediction-as/) defines a task in which all four sentiment elements must be produced at once, and solves it by generating paraphrases. In 2023, Alibaba Group and DAMO Academy share [Sentiment Analysis in the Era of Large Language Models: A Reality Check](https://md.research.elamur.ai/free/en/papers/zhang-2023-sentiment-analysis-era-large-language/). It weighs how well large models handle sentiment tasks, faults present-day testing for uneven prompting and for covering too few kinds of task, and releases resources for fairer comparison.

Reasoning, dialogue and reading comprehension are the hallmark of Westlake University among the works here. LogiQA is a reading comprehension challenge built around logical reasoning rather than multi-hop evidence or commonsense. MuTual is a hand-annotated dataset of multi-turn dialogues designed to test reasoning in conversation. DialogSum gathers dialogues from real-life settings as a testbed for summarising conversations. In 2023 the university extends this line to [Evaluating the Logical Reasoning Ability of ChatGPT and GPT-4](https://md.research.elamur.ai/free/en/papers/liu-2023-evaluating-logical-reasoning-ability-chatgpt/). To check robustness, that study builds a fresh out-of-distribution test set from Chinese civil servant examination questions of 2022.

Evaluation and robustness of language models form a strong common thread across institutions. Westlake University's [A Survey on Evaluation of Large Language Models](https://md.research.elamur.ai/free/en/papers/chang-2023-survey-evaluation-large-language-models/) maps what to evaluate, where and how. [PromptRobust](https://md.research.elamur.ai/free/en/papers/zhu-2023-promptrobust-towards-evaluating-robustness-large/) adapts black-box textual attacks to prompts, so that one can measure how sensitive models are to small changes in instructions. [PandaLM](https://md.research.elamur.ai/free/en/papers/wang-2023-pandalm-automatic-evaluation-benchmark-llm/) starts from the habit of reusing Alpaca's hyperparameters when tuning open models. It offers an automatic judge so that tuning choices can be compared more cheaply. At Zhejiang University, [MMBench](https://md.research.elamur.ai/free/en/papers/liu-2023-mmbench-multi-modal-model-all/) takes up the hard problem of evaluating vision-language models thoroughly. Alibaba Group's [Text-to-SQL Empowered by Large Language Models: A Benchmark Evaluation](https://md.research.elamur.ai/free/en/papers/gao-2023-text-sql-empowered-large-language/) compares prompting strategies systematically. From those findings it builds DAIL-SQL, a method that topped the Spider leaderboard. The robustness concern first seen in TextBugger thus returns in a new form.

Training and adapting language models is another line, mostly carried by Alibaba and its partners. [Instruction Tuning for Large Language Models: A Survey](https://md.research.elamur.ai/free/en/papers/zhang-2023-instruction-tuning-large-language-models/), shared by Alibaba Group and Zhejiang University, starts from the gap between next-token prediction and what users actually want from a model. It reviews the methods, datasets and models that bridge that gap. [LLM-Adapters](https://md.research.elamur.ai/free/en/papers/hu-2023-llm-adapters-adapter-family-parameter/), shared by Alibaba Group and DAMO Academy, studies parameter-efficient adapters. It gives practical advice on how to design them, together with a toolkit others can reuse, so that large models can be adapted at modest cost. Zhejiang Lab's [Unifying Large Language Models and Knowledge Graphs: A Roadmap](https://md.research.elamur.ai/free/en/papers/pan-2023-unifying-large-language-models-knowledge/) sets out three frameworks for joining models with structured knowledge. The lab's [Chinese Idiom Paraphrasing](https://md.research.elamur.ai/free/en/papers/qiang-2023-chinese-idiom-paraphrasing/) defines a new task of rewriting idioms in plain Chinese. The Dice loss work at Zhejiang University belongs to the same broad concern with training objectives. It replaces cross-entropy with losses based on the Sørensen–Dice coefficient to cope with imbalanced labels.

Multimodal models are a large and growing field in the region. Alibaba Group's [Qwen-VL](https://md.research.elamur.ai/free/en/papers/bai-2023-qwen-vl-versatile-vision-language/) builds a family of vision-language models on top of Qwen-7B. They cover image understanding, the localisation of objects and the reading of text in images. DAMO Academy's [Video-LLaMA](https://md.research.elamur.ai/free/en/papers/zhang-2023-video-llama-instruction-tuned-audio/) joins both visual and audio streams to a language model for video understanding, at a time when most multimodal models linked a single extra modality. DAMO Academy's [study of object hallucinations in vision-language models](https://md.research.elamur.ai/free/en/papers/leng-2023-mitigating-object-hallucinations-large-vision/) looks at why such models describe objects that are not there. It shows that degrading the clarity of the image strengthens the underlying tendencies, and uses that finding to reduce hallucinated objects. DeepSeek adds [DeepSeek-VL](https://md.research.elamur.ai/free/en/papers/lu-2024-deepseek-vl-real-world/), an open vision-language model aimed at real-world use. Together with ActivityNet-QA and MMBench, these works make vision and language one of the region's broadest strengths on this site.

Mathematical and formal reasoning is the most cited line among the works here, and DeepSeek leads it. [DeepSeekMath](https://md.research.elamur.ai/free/en/papers/shao-2024-deepseekmath-pushing-limits-mathematical-reasoning/) begins from the gap between open models and closed systems on competition-level mathematics and works to narrow it. [Math-Shepherd](https://md.research.elamur.ai/free/en/papers/wang-2024-math-shepherd-verify-reinforce-llms/) trains a process reward model on step-level labels produced automatically rather than by people. It uses that model both to check solutions and to guide reinforcement learning. [DeepSeek-R1](https://md.research.elamur.ai/free/en/papers/deepseek-ai-2025-deepseek-r1-incentivizing-reasoning-capability/) argues that reasoning ability can arise from reinforcement learning on its own. [DeepSeek-Prover-V2](https://md.research.elamur.ai/free/en/papers/ren-2025-deepseek-prover-v2-advancing-formal/) carries the line into formal theorem proving, building its training data through a recursive pipeline.

Code and applied systems round out the profile. Alibaba Group's [Qwen2.5-Coder Technical Report](https://md.research.elamur.ai/free/en/papers/hui-2024-qwen2-5-coder-technical-report/) presents a family of open code models, written against the backdrop of open code models trailing proprietary systems. Hangzhou Dianzi University's [ChatDoctor](https://md.research.elamur.ai/free/en/papers/li-2023-chatdoctor-medical-chat-model-fine/) builds a medical assistant on LLaMA-7B, trained in two stages. Ant Group's [A-MEM](https://md.research.elamur.ai/free/en/papers/xu-2025-mem-agentic-memory-llm-agents/) proposes a more flexible memory for language-model agents, in contrast to designs with fixed schemas and hard-coded read and write points. Ant Group's [Large Language Diffusion Models](https://md.research.elamur.ai/free/en/papers/nie-2025-large-language-diffusion-models/) trains a masked diffusion model, LLaDA, from scratch. It tests whether the abilities of large language models really depend on generating text one token at a time from left to right.

### Leading institutions and labs

Alibaba Group is the region's largest and most cited institution on this site. Its works run from adversarial text in 2018 and aspect-based sentiment in 2020 to the instruction-tuning survey, text-to-SQL evaluation and LLM-Adapters in 2023. In that same year it released Qwen-VL, and in 2024 it published Qwen2.5-Coder. Its research moves from analysing text toward building its own open model families for vision and for code.

Zhejiang University is the region's largest academic contributor here and its oldest, with the 2011 opinion-mining paper. Its works centre on named entity recognition and training objectives, then turn to multimodal evaluation with MMBench and to entity recognition with large models in GPT-NER. It shares several works with Alibaba Group and with Westlake University.

Westlake University has a distinct emphasis on reasoning, dialogue and evaluation. Its datasets of 2020 and 2021 lead into its surveys and benchmarks of 2023 on how language models should be judged and how robust they are.

DeepSeek has a narrow but highly cited body of work. It covers mathematical reasoning, process reward models, vision-language understanding, reinforcement learning for reasoning and formal proof, and it forms the most concentrated research line in the region.

Alibaba DAMO Academy carries the finer-grained side of information extraction and sentiment analysis. It also contributes multimodal work on video and on hallucination.

Hangzhou Dianzi University appears through video question answering and a medical chat model. Zhejiang Lab, the region's research institute here, contributes the roadmap on knowledge graphs and language models and the idiom paraphrasing task. Ant Group's two works from 2025 point to the newest directions in the region: agent memory and diffusion language models.

### Leading scientists

The people come here institution by institution, in the order each institution's own presentation gives them.

At Alibaba Group, the first names are [Jingren Zhou](https://md.research.elamur.ai/free/en/authors/jingren-zhou/), [Binyuan Hui](https://md.research.elamur.ai/free/en/authors/binyuan-hui/), [Fei Huang](https://md.research.elamur.ai/free/en/authors/fei-huang/), [Bowen Yu](https://md.research.elamur.ai/free/en/authors/bowen-yu/), Rui Men, [Junyang Lin](https://md.research.elamur.ai/free/en/authors/junyang-lin/), [Shuai Bai](https://md.research.elamur.ai/free/en/authors/shuai-bai/) and Jianhong Tu. Their institution's research here runs through open model families such as Qwen-VL and Qwen2.5-Coder, as well as earlier work on sentiment and robustness. Jingren Zhou is also among the people of Ant Group, which makes Zhou one of the personal links between the region's companies.

At Zhejiang University, the people begin with [Jiwei Li](https://md.research.elamur.ai/free/en/authors/jiwei-li/), [Fei Wu](https://md.research.elamur.ai/free/en/authors/fei-wu/), [Xiaoya Li](https://md.research.elamur.ai/free/en/authors/xiaoya-li/), [Cunxiang Wang](https://md.research.elamur.ai/free/en/authors/cunxiang-wang/), Xiaofei Sun, Junbo Zhao, Yuxian Meng and [Yueting Zhuang](https://md.research.elamur.ai/free/en/authors/yueting-zhuang/). The university's work here centres on entity recognition framed as reading comprehension, Dice-based training losses, and entity recognition with large models, along with multimodal evaluation. Cunxiang Wang is also among the people of Westlake University, a direct personal tie between the two universities.

At Westlake University, the people begin with [Yue Zhang](https://md.research.elamur.ai/free/en/authors/yue-zhang/), [Linyi Yang](https://md.research.elamur.ai/free/en/authors/linyi-yang/), Cunxiang Wang, [Leyang Cui](https://md.research.elamur.ai/free/en/authors/leyang-cui/), Yidong Wang, Yulong Chen, Zhenzhong Lan and [Hanmeng Liu](https://md.research.elamur.ai/free/en/authors/hanmeng-liu/). Their university's research here covers logical reasoning, dialogue reasoning and summarisation datasets, the evaluation survey, robustness of prompts and automatic judging of instruction tuning.

At DeepSeek, the people begin with Junxiao Song, [Daya Guo](https://md.research.elamur.ai/free/en/authors/daya-guo/), Chong Ruan, [Zhihong Shao](https://md.research.elamur.ai/free/en/authors/zhihong-shao/), Runxin Xu, Chengqi Deng, Zhenda Xie and Huazuo Gao. They belong to the region's reasoning line, from DeepSeekMath and Math-Shepherd to DeepSeek-R1 and DeepSeek-Prover-V2.

At Alibaba DAMO Academy, the first names are [Lidong Bing](https://md.research.elamur.ai/free/en/authors/lidong-bing/), [An Yang](https://md.research.elamur.ai/free/en/authors/an-yang/), Xin Li, [Wenxuan Zhang](https://md.research.elamur.ai/free/en/authors/wenxuan-zhang/), Haiyang Xu, Zhirui Zhang, Nguyen Bach and Ruidan He. The academy's research here spans structured sentiment extraction, joint information extraction, ChatIE, Video-LLaMA and the study of hallucinations in vision-language models.

Hangzhou Dianzi University's people begin with Zihan Li, Rui Zhang, Jingjing Liu, Lijuan Wang, Hao Fang, Yiming Cui, Jianfeng Lu and Zifan Wang. The university's works here are on video question answering and medical dialogue.

At Ant Group, the people begin with Bo Zheng, Chen Li, Jingren Zhou, Jingdong Chen, Haitao Mi, Chengwei Qin, Wenqiang Lei and Xibin Wu. They are connected to the work on agent memory and diffusion language models.

At Zhejiang Lab, the first name is [Xindong Wu](https://md.research.elamur.ai/free/en/authors/xindong-wu/), followed by Jiaxi Yang, Zifan Wang, Hongyu Lin, Chen Zhang, Wenbin Jiang and Liqiang Wang. The lab's research here joins language models to knowledge graphs and studies Chinese idiom paraphrasing. Zifan Wang appears both here and at Hangzhou Dianzi University.

### Landmark works

The most cited Hangzhou work in our library is DeepSeek-R1 from DeepSeek. Its central claim is that strong reasoning in language models can be drawn out through reinforcement learning without relying on supervised examples of reasoning. That claim has made it a reference point for much later work collected here.

DeepSeekMath, the next most cited, set out to bring open models closer to closed systems on hard mathematics problems. It is one of the foundations of the region's reasoning line.

Qwen-VL from Alibaba Group is the region's most cited multimodal work. It extends a language model to understand images, locate objects within them and read the text they contain.

Westlake University's survey on evaluating large language models is the most cited work from the region's universities on this site. It organises a fast-growing body of evaluation methods around the questions of what to test, where, and how.

The Qwen2.5-Coder report and Zhejiang University's MMBench follow among the most cited. The first describes open models for code; the second offers a systematic benchmark for vision-language models. Math-Shepherd stands out for showing that step-by-step supervision of reasoning can be built without human annotation. The instruction-tuning survey, shared by Alibaba Group and Zhejiang University, is a widely cited overview of how pretrained models are made to follow instructions.

Among earlier works, LogiQA, ONEIE and TextBugger draw steady citation here. Each comes from a different institution, which shows that the region's influence on this site is not confined to the era of large models.

### Collaborations

Several works in our library join institutions within Hangzhou, and these joint works show how its industry and universities are linked.

The link between Alibaba Group and Zhejiang University is the oldest such tie here. It runs from TextBugger in 2018 to the instruction-tuning survey in 2023, so the region's largest company and largest university have worked together across both the earlier and the later phases of its research.

Within the Alibaba family, Alibaba Group and Alibaba DAMO Academy share the sentiment reality check and LLM-Adapters. The two thus co-author on both applied analysis and model adaptation.

Between universities, Zhejiang University and Westlake University share the template-based entity recognition work. Cunxiang Wang, who appears at both, adds a personal tie between the two.

Other individuals bridge institutions too. Jingren Zhou appears at both Alibaba Group and Ant Group. Zifan Wang appears at both Hangzhou Dianzi University and Zhejiang Lab. These overlaps suggest a local research community in which people move between companies, universities and the research institute, or hold ties to more than one.

Outside the region, all eight institutions are placed in China. The works themselves reach well beyond: they engage with models and benchmarks from elsewhere, building on LLaMA, testing ChatGPT and GPT-4, and competing on the Spider leaderboard.

### Timeline

In 2011, the earliest work from the region on this site appears: Zhejiang University's double-propagation method for opinion words and targets. It opens the long line of sentiment and opinion research.

Between 2018 and 2019, Alibaba Group and Zhejiang University share TextBugger on adversarial text, and Hangzhou Dianzi University publishes ActivityNet-QA for video question answering. Robustness and video understanding thus join opinion mining among the region's topics.

In 2020, core language processing reaches its peak. Zhejiang University's entity recognition framework and Dice loss appear, alongside Alibaba DAMO Academy's ONEIE and triplet extraction, Alibaba Group's graph attention model for aspect sentiment, and Westlake University's LogiQA and MuTual.

In 2021, the earlier lines continue with DialogSum, few-shot entity recognition with templates, and aspect sentiment quad prediction. These works carry dialogue, extraction and sentiment research forward just before the shift to large language models.

The year 2023 is the broadest in the region's output here. Surveys on instruction tuning, evaluation, and the joining of models with knowledge graphs appear in this year. So do benchmarks such as MMBench, PromptRobust and PandaLM; models such as Qwen-VL, Video-LLaMA and ChatDoctor; and studies that test sentiment analysis, extraction, logical reasoning and text-to-SQL with large models.

In 2024, DeepSeek's DeepSeekMath, Math-Shepherd and DeepSeek-VL appear, together with Alibaba Group's Qwen2.5-Coder. Mathematical reasoning and open model families become central.

In 2025, DeepSeek-R1 and DeepSeek-Prover-V2 extend the reasoning line to reinforcement learning and formal proof. Ant Group's A-MEM and LLaDA open work on agent memory and diffusion-based language models, and Zhejiang University's GPT-NER links its long tradition in entity recognition to large language models. Among the works here, this is the most recent stage of the Hangzhou region's research.

## Institutions (8)

3 universities, 1 research institute, 4 companies.

### 🎓 Universities (3)

1. **[Zhejiang University](https://md.research.elamur.ai/free/en/universities/zhejiang-university/)**University · Hangzhou, [China](https://md.research.elamur.ai/free/en/countries/china/)*33 works · 10,088 citations*
2. **[Westlake University](https://md.research.elamur.ai/free/en/universities/westlake-university/)**University · Hangzhou, [China](https://md.research.elamur.ai/free/en/countries/china/)*22 works · 6,635 citations*
3. **[Hangzhou Dianzi University](https://md.research.elamur.ai/free/en/universities/hangzhou-dianzi-university/)**University · Hangzhou, [China](https://md.research.elamur.ai/free/en/countries/china/)*2 works · 1,029 citations*

### 🔬 Labs and research institutes (1)

1. **[Zhejiang Lab](https://md.research.elamur.ai/free/en/labs/zhejiang-lab/)**Research institute · Hangzhou, [China](https://md.research.elamur.ai/free/en/countries/china/)*2 works · 1,576 citations*

### 🏢 Companies (4)

1. **[Alibaba Group](https://md.research.elamur.ai/free/en/companies/alibaba-group/)**Company · Hangzhou, [China](https://md.research.elamur.ai/free/en/countries/china/)*39 works · 10,954 citations*
2. **[DeepSeek](https://md.research.elamur.ai/free/en/companies/deepseek/)**Company · Hangzhou, [China](https://md.research.elamur.ai/free/en/countries/china/)*5 works · 7,409 citations*
3. **[Alibaba DAMO Academy](https://md.research.elamur.ai/free/en/companies/alibaba-damo-academy/)**Company · Hangzhou, [China](https://md.research.elamur.ai/free/en/countries/china/)*14 works · 3,843 citations*
4. **[Ant Group](https://md.research.elamur.ai/free/en/companies/ant-group/)**Company · Hangzhou, [China](https://md.research.elamur.ai/free/en/countries/china/)*2 works · 335 citations*

## Works in our library (92)

Works of its institutions, the most cited first.

1. [**DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning**DeepSeek-AI, Daya Guo, Dejian Yang, et al.*2025 · 4,593 citations*](https://md.research.elamur.ai/free/en/papers/deepseek-ai-2025-deepseek-r1-incentivizing-reasoning-capability/)
2. [**A Survey on Evaluation of Large Language Models**Yupeng Chang, Xu Wang, Jindong Wang, et al.*2023 · 3,640 citations*](https://md.research.elamur.ai/free/en/papers/chang-2023-survey-evaluation-large-language-models/)
3. [**DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models**Zhihong Shao, Peiyi Wang, Qihao Zhu, et al.*2024 · 2,013 citations*](https://md.research.elamur.ai/free/en/papers/shao-2024-deepseekmath-pushing-limits-mathematical-reasoning/)
4. [**Unifying Large Language Models and Knowledge Graphs: A Roadmap**Shirui Pan, Linhao Luo, Yufei Wang, et al.*2023 · 1,545 citations*](https://md.research.elamur.ai/free/en/papers/pan-2023-unifying-large-language-models-knowledge/)
5. [**Opinion Word Expansion and Target Extraction through Double Propagation**Guang Qiu, Bing Liu, Jiajun Bu, et al.*2011 · 1,196 citations*](https://md.research.elamur.ai/free/en/papers/qiu-2011-opinion-word-expansion-target-extraction/)
6. [**Qwen-VL: A Versatile Vision-Language Model for Understanding, Localization, Text Reading, and Beyond**Jinze Bai, Shuai Bai, Shusheng Yang, et al.*2023 · 913 citations*](https://md.research.elamur.ai/free/en/papers/bai-2023-qwen-vl-versatile-vision-language/)
7. [**MMBench: Is Your Multi-modal Model an All-around Player?**Yuan Liu, Haodong Duan, Yuanhan Zhang, et al.*2023 · 907 citations*](https://md.research.elamur.ai/free/en/papers/liu-2023-mmbench-multi-modal-model-all/)
8. [**A Unified MRC Framework for Named Entity Recognition**Xiaoya Li, Jingrong Feng, Yuxian Meng, et al.*2020 · 833 citations*](https://md.research.elamur.ai/free/en/papers/li-2020-unified-mrc-framework-named-entity/)
9. [**Relational Graph Attention Network for Aspect-based Sentiment Analysis**Kai Wang, Weizhou Shen, Yunyi Yang, et al.*2020 · 775 citations*](https://md.research.elamur.ai/free/en/papers/wang-2020-relational-graph-attention-network-aspect/)
10. [**Dice Loss for Data-imbalanced NLP Tasks**Xiaoya Li, Xiaofei Sun, Yuxian Meng, et al.*2020 · 734 citations*](https://md.research.elamur.ai/free/en/papers/li-2020-dice-loss-data-imbalanced-nlp/)
11. [**ChatDoctor: A Medical Chat Model Fine-Tuned on a Large Language Model Meta-AI (LLaMA) Using Medical Domain Knowledge**Yunxiang Li, Zihan Li, Kai Zhang, et al.*2023 · 640 citations*](https://md.research.elamur.ai/free/en/papers/li-2023-chatdoctor-medical-chat-model-fine/)
12. [**A Joint Neural Model for Information Extraction with Global Features**Ying Lin, Heng Ji, Fei Huang, et al.*2020 · 600 citations*](https://md.research.elamur.ai/free/en/papers/lin-2020-joint-neural-model-information-extraction/)
13. [**Qwen2.5-Coder Technical Report**Binyuan Hui, Jian Yang, Zeyu Cui, et al.*2024 · 547 citations*](https://md.research.elamur.ai/free/en/papers/hui-2024-qwen2-5-coder-technical-report/)
14. [**TextBugger: Generating Adversarial Text Against Real-world Applications**Jinfeng Li, Shouling Ji, Tianyu Du, et al.*2018 · 541 citations*](https://md.research.elamur.ai/free/en/papers/li-2018-textbugger-generating-adversarial-text-against/)
15. [**Instruction Tuning for Large Language Models: A Survey**Shengyu Zhang, Linfeng Dong, Xiaoya Li, et al.*2023 · 515 citations*](https://md.research.elamur.ai/free/en/papers/zhang-2023-instruction-tuning-large-language-models/)
16. [**Sentiment Analysis in the Era of Large Language Models: A Reality Check**Wenxuan Zhang, Yue Deng, Bing Liu, et al.*2023 · 505 citations*](https://md.research.elamur.ai/free/en/papers/zhang-2023-sentiment-analysis-era-large-language/)
17. [**Math-Shepherd: Verify and Reinforce LLMs Step-by-step without Human Annotations**Peiyi Wang, Lei Li, Zhihong Shao, et al.*2024 · 482 citations*](https://md.research.elamur.ai/free/en/papers/wang-2024-math-shepherd-verify-reinforce-llms/)
18. [**Text-to-SQL Empowered by Large Language Models: A Benchmark Evaluation**Dawei Gao, Haibin Wang, Yaliang Li, et al.*2023 · 461 citations*](https://md.research.elamur.ai/free/en/papers/gao-2023-text-sql-empowered-large-language/)
19. [**GPT-NER: Named Entity Recognition via Large Language Models**Shuhe Wang, Xiaofei Sun, Xiaoya Li, et al.*2025 · 436 citations*](https://md.research.elamur.ai/free/en/papers/wang-2025-gpt-ner-named-entity-recognition/)
20. [**Template-Based Named Entity Recognition Using BART**Leyang Cui, Yu Wu, Jian Liu, et al.*2021 · 420 citations*](https://md.research.elamur.ai/free/en/papers/cui-2021-template-based-named-entity-recognition/)
21. [**LLM-Adapters: An Adapter Family for Parameter-Efficient Fine-Tuning of Large Language Models**Zhiqiang Hu, Lei Wang, Yihuai Lan, et al.*2023 · 398 citations*](https://md.research.elamur.ai/free/en/papers/hu-2023-llm-adapters-adapter-family-parameter/)
22. [**LogiQA: A Challenge Dataset for Machine Reading Comprehension with Logical Reasoning**Jian Liu, Leyang Cui, Hanmeng Liu, et al.*2020 · 395 citations*](https://md.research.elamur.ai/free/en/papers/liu-2020-logiqa-challenge-dataset-machine-reading/)
23. [**ActivityNet-QA: A Dataset for Understanding Complex Web Videos via Question Answering**Zhou Yu, Dejing Xu, Jun Yu, et al.*2019 · 389 citations*](https://md.research.elamur.ai/free/en/papers/yu-2019-activitynet-qa-dataset-understanding-complex/)
24. [**Position-Aware Tagging for Aspect Sentiment Triplet Extraction**Lu Xu, Hao Li, Wei Lu, et al.*2020 · 370 citations*](https://md.research.elamur.ai/free/en/papers/xu-2020-position-aware-tagging-aspect-sentiment/)
25. [**Persuasion for Good: Towards a Personalized Persuasive Dialogue System for Social Good**Xuewei Wang, Weiyan Shi, Richard Kim, et al.*2019 · 366 citations*](https://md.research.elamur.ai/free/en/papers/wang-2019-persuasion-good-personalized-persuasive-dialogue/)

[See all 92 works →](https://md.research.elamur.ai/free/en/agglomerations/hangzhou/works/)

## Works by year

### 2020s

[**2026** 2 works](https://md.research.elamur.ai/free/en/agglomerations/hangzhou/year/2026/) · [**2025** 10 works](https://md.research.elamur.ai/free/en/agglomerations/hangzhou/year/2025/) · [**2024** 17 works](https://md.research.elamur.ai/free/en/agglomerations/hangzhou/year/2024/) · [**2023** 31 works](https://md.research.elamur.ai/free/en/agglomerations/hangzhou/year/2023/) · [**2022** 7 works](https://md.research.elamur.ai/free/en/agglomerations/hangzhou/year/2022/) · [**2021** 6 works](https://md.research.elamur.ai/free/en/agglomerations/hangzhou/year/2021/) · [**2020** 13 works](https://md.research.elamur.ai/free/en/agglomerations/hangzhou/year/2020/)

### 2010s

[**2019** 4 works](https://md.research.elamur.ai/free/en/agglomerations/hangzhou/year/2019/) · [**2018** 1 work](https://md.research.elamur.ai/free/en/papers/li-2018-textbugger-generating-adversarial-text-against/) · [**2011** 1 work](https://md.research.elamur.ai/free/en/papers/qiu-2011-opinion-word-expansion-target-extraction/)

## Concepts, shelves and topics

Of its works in our library: the number of works on each and its rating among them.

💡 Concepts 4[Named entity recognition · 3 works](https://md.research.elamur.ai/free/en/concepts/named-entity-recognition/) · [Language model · 1 work](https://md.research.elamur.ai/free/en/concepts/language-model/) · [Semantic parsing · 1 work](https://md.research.elamur.ai/free/en/concepts/semantic-parsing/) · [Context-free grammar · 1 work](https://md.research.elamur.ai/free/en/concepts/context-free-grammar/)

📚 Shelves 10[Natural language processing · 43 works](https://md.research.elamur.ai/free/en/shelves/nlp/) · [Other · 24 works](https://md.research.elamur.ai/free/en/shelves/other/) · [Mathematics & statistics · 8 works](https://md.research.elamur.ai/free/en/shelves/mathematics/) · [AI & machine learning · 8 works](https://md.research.elamur.ai/free/en/shelves/ai/) · [Linguistics & languages · 14 works](https://md.research.elamur.ai/free/en/shelves/linguistics/) · [Machine translation · 6 works](https://md.research.elamur.ai/free/en/shelves/machine-translation/) · [Machine translation · 2 works](https://md.research.elamur.ai/free/en/shelves/mtrans/) · [Aramaic · 1 work](https://md.research.elamur.ai/free/en/shelves/arc/) · [Law · 1 work](https://md.research.elamur.ai/free/en/shelves/law/) · [English · 1 work](https://md.research.elamur.ai/free/en/shelves/en/)

🧭 Topics 26[Language models & pretraining · 39 works](https://md.research.elamur.ai/free/en/topics/language-models/) · [Reasoning · 14 works](https://md.research.elamur.ai/free/en/topics/reasoning/) · [Vision & language · 8 works](https://md.research.elamur.ai/free/en/topics/multimodal/) · [Named entity recognition · 5 works](https://md.research.elamur.ai/free/en/topics/named-entity-recognition/) · [Sentiment & emotion · 7 works](https://md.research.elamur.ai/free/en/topics/sentiment-emotion/) · [Evaluation & benchmarks · 9 works](https://md.research.elamur.ai/free/en/topics/evaluation-benchmarks/) · [Information extraction · 4 works](https://md.research.elamur.ai/free/en/topics/information-extraction/) · [Dialogue systems · 5 works](https://md.research.elamur.ai/free/en/topics/dialogue-systems/) · [Question answering · 2 works](https://md.research.elamur.ai/free/en/topics/question-answering/) · [Coreference resolution · 4 works](https://md.research.elamur.ai/free/en/topics/coreference-resolution/) · [Information retrieval & search · 3 works](https://md.research.elamur.ai/free/en/topics/information-retrieval/) · [Knowledge graphs & ontologies · 1 work](https://md.research.elamur.ai/free/en/topics/knowledge-graphs/) · [Semantic parsing & meaning representation · 4 works](https://md.research.elamur.ai/free/en/topics/semantic-parsing/) · [Speech & spoken language · 2 works](https://md.research.elamur.ai/free/en/topics/speech/) · [Machine translation · 4 works](https://md.research.elamur.ai/free/en/topics/machine-translation/)+11 more

Agglomerations: the GHSL Urban Centre Database 2025; populations: the GHSL functional urban areas 2015 (urban centre and commuting zone), else the urban centre, as each page says (European Commission, Joint Research Centre, CC BY 4.0). An institution belongs to the urban centre its coordinates (ROR, OpenAlex) fall in, or one within 5 km. Land: Natural Earth.

- [← back to](https://md.research.elamur.ai/free/en/countries/china/) — China

Updated 23 October 2026