开源社区中项目热度阶段预测方法研究——以大语言模型项目为例

卢小艺; 陈阳; 傅晓明; 李聪

doi:10.11959/j.issn.2096-6652.202542

您当前的位置：

首页 >

文章列表页 >

开源社区中项目热度阶段预测方法研究——以大语言模型项目为例

专题：社会计算与韧性社会 | 更新时间：2026-03-18

- 开源社区中项目热度阶段预测方法研究——以大语言模型项目为例
- Research on popularity stage prediction methods for open-source community projects: a case study of large language model projects
- 智能科学与技术学报 2025年7卷第4期页码：505-516
- 作者机构：
  
  1.复旦大学计算与智能创新学院，上海 200438
  2.复旦大学未来信息创新学院自适应网络与控制研究室，上海 200438
  3.复旦大学社会智能研究中心，上海 200438
  4.哥廷根大学计算机科学研究所，哥廷根 37073
- 作者简介：
  
  [ "卢小艺（2003- ），女，复旦大学计算与智能创新学院硕士生，主要研究方向为复杂网络理论及应用、人工智能驱动的网络科学研究。" ]
  [ "陈阳（1981- ），男，博士，复旦大学计算与智能创新学院教授、博士生导师，上海市智能信息处理重点实验室副主任，主要研究方向为社会计算、计算机网络、大语言模型。" ]
  [ "傅晓明（1973- ），男，博士，哥廷根大学计算机科学研究所教授，复旦大学社会智能研究中心首席科学家，德国国家工程院院士，欧洲科学院院士，IEEE Fellow、ACM杰出科学家，主要研究方向为社会计算、计算机网络等。" ]
  [ "李聪（1986- ），女，博士，复旦大学未来信息创新学院副研究员，主要研究方向为复杂网络理论及应用、人类行为分析、人工智能驱动的网络科学。" ]
- 基金信息：
  
  国家自然科学基金项目(62173095;U23A20331;62573133;62072115);上海市科技创新行动计划项目(22510713600)
- DOI：10.11959/j.issn.2096-6652.202542
  中图分类号： TP181
- 收稿：2025-09-26，
  
  修回：2025-12-02，
  
  录用：2025-12-04，
  
  纸质出版：2025-12-15
- 稿件说明：
移动端阅览
卢小艺,陈阳,傅晓明等.开源社区中项目热度阶段预测方法研究——以大语言模型项目为例[J].智能科学与技术学报,2025,07(04):505-516.

LU Xiaoyi,CHEN Yang,FU Xiaoming,et al.Research on popularity stage prediction methods for open-source community projects: a case study of large language model projects[J].Chinese Journal of Intelligent Science and Technology,2025,07(04):505-516.
卢小艺,陈阳,傅晓明等.开源社区中项目热度阶段预测方法研究——以大语言模型项目为例[J].智能科学与技术学报,2025,07(04):505-516. DOI： 10.11959/j.issn.2096-6652.202542.

LU Xiaoyi,CHEN Yang,FU Xiaoming,et al.Research on popularity stage prediction methods for open-source community projects: a case study of large language model projects[J].Chinese Journal of Intelligent Science and Technology,2025,07(04):505-516. DOI： 10.11959/j.issn.2096-6652.202542.

摘要

近些年，GitHub开源社区的累计注册开发者数量已超过1亿，不同项目在用户关注度上的竞争日益激烈。理解开源社区中项目热度变化不仅有助于研究者和开发者把握技术趋势，还能为资源配置、合作选择和科研投入提供参考。然而，现有研究多将热度视为关注度的数值回归或流行度二分类问题，忽略了热度随时间演化所呈现的阶段性特征及其模糊过渡。为此，以开源大语言模型项目为研究对象，构建了热度阶段划分与预测一体化框架，实现了热度阶段的自动刻画与预测。具体而言，先在固定窗口内提取项目关注度增长速率与总量构建特征向量，采用高斯混合模型进行软聚类，获得4类可解释的热度阶段及其概率标签；随后，设计基于时间融合变换器（temporal fusion transformer，TFT）的预测模型，利用包含关注度、分支数、提交数等五维行为指标的序列预测未来窗口内的阶段概率分布。实验结果表明，该方法在交叉熵、均方误差、平均绝对误差和准确率等指标上优于已有方法，并在4类热度演化模式中均保持较高精度，验证了其泛化能力。该方法为开源项目提供了热度阶段预测框架，具有良好的实用与推广价值。

Abstract

In recent years

the cumulative number of registered developers on the open-source community GitHub has exceeded 100 million

leading to increasingly fierce competition for user attention among projects. Understanding the dynamics of project popularity in open-source communities not only helps researchers and developers grasp technological trends

but also provides valuable references for resource allocation

collaboration decisions

and research investment. However

existing studies often treat popularity as a continuous numerical regression or binary classification task

overlooking the stage-wise evolutionary patterns and gradual transitions exhibited in real project popularity dynamics. Open-source large language model (LLM) projects were taken as the research object. An integrated framework for popularity stage segmentation and prediction was proposed

enabling the automatic characterization and forecasting of project popularity stages. Specifically

feature vectors based on growth rate and cumulative scale of project attention within fixed windows were extracted

and a Gaussian mixture model (GMM) for soft clustering was employed to obtain four interpretable stages of popularity with probabilistic labels. Subsequently

a prediction model based on the temporal fusion transformer (TFT) was designed

which leveraged sequential data of five behavioral indicators

such as attention

number of forks

and number of commits

to forecast the stage probability distribution in future windows. Experimental results demonstrate that the proposed method outperforms existing baselines in terms of cross-entropy

mean squared error

mean absolute error

and accuracy

while maintaining high precision across all four patterns of popularity evolution

thereby validating its generalization capability. The proposed approach provides a practical and scalable framework for popularity stage prediction of open-source projects

offering strong utility and potential for applications.

关键词

Keywords

references

HAN J X, DENG S G, XIA X, et al. Characterization and prediction of popular projects on GitHub[C]//Proceedings of the 2019 IEEE 43rd Annual Computer Software and Applications Conference (COMPSAC). Piscataway: IEEE Press, 2019: 21-26.

FAN Y R, XIA X, LO D, et al. What makes a popular academic AI repository?[J]. Empirical Software Engineering, 2021, 26(1): 2.

LIM B, ARıK S Ö, LOEFF N, et al. Temporal fusion transformers for interpretable multi-horizon time series forecasting[J]. International Journal of Forecasting, 2021, 37(4): 1748-1764.

SAHIN S E, KARPAT K, TOSUN A. Predicting popularity of open source projects using recurrent neural networks[C]//Proceedings of the 2019 Open Source Systems. Cham: Springer, 2019: 80-90.

BORGES H, HORA A, VALENTE M T. Understanding the factors that impact the popularity of GitHub repositories[C]//Proceedings of the 2016 IEEE International Conference on Software Maintenance and Evolution (ICSME). Piscataway: IEEE Press, 2016: 334-344.

BORGES H, HORA A, VALENTE M T. Predicting the popularity of GitHub repositories[C]//Proceedings of the 12th International Conference on Predictive Models and Data Analytics in Software Engineering. New York: ACM Press, 2016: 1-10.

VARUNA T V, MOHAN A. Trend prediction of GitHub using time series analysis[C]//Proceedings of the 2019 10th International Conference on Computing, Communication and Networking Technologies (ICCCNT). Piscataway: IEEE Press, 2019: 1-7.

何玉林, 杨锦, 黄哲学, 等. 基于标签迭代的聚类集成算法[J]. 智能科学与技术学报, 2024, 6(4): 466-479.

HE Y L, YANG J, HUANG Z X, et al. Label iteration-based clustering ensemble algorithm[J]. Chinese Journal of Intelligent Science and Technology, 2024, 6(4): 466-479.

胡毅, 瞿博阳, 梁静, 等. 进化集成学习算法综述[J]. 智能科学与技术学报, 2021, 3(1): 18-33.

HU Y, QU B Y, LIANG J, et al. A survey on evolutionary ensemble learning algorithm[J]. Chinese Journal of Intelligent Science and Technology, 2021, 3(1): 18-33.

丁宸聪. 基于选择性深度嵌入聚类的复杂未知雷达信号分选方法[J]. 智能科学与技术学报, 2025, 7(3): 329-337.

DING C C. A complex unknown radar signal deinterleaving method based on selective deep embedding clustering[J]. Chinese Journal of Intelligent Science and Technology, 2025, 7(3): 329-337.

LI C, SHI M, QU B, et al. Deep attributed network representation learning via attribute enhanced neighborhood[J]. Neurocomputing, 2022, 508: 170-181.

BONFANTI M. A framework for identifying evolution patterns of open-source software projects[D]. Delft: Delft University of Technology, 2021.

PICKERILL P, JUNGEN H J, OCHODEK M, et al. PHANTOM: curating GitHub for engineered software projects using time-series clustering[J]. Empirical Software Engineering, 2020, 25(4): 2897-2929.

LIU S Z, YAO S C, LIU D X, et al. A latent hawkes process model for event clustering and temporal dynamics learning with applications in GitHub[C]//Proceedings of the 2019 IEEE 39th International Conference on Distributed Computing Systems (ICDCS). Piscataway: IEEE Press, 2019: 1275-1285.

ZHOU Y R, QIAO Y, XU T. Exploring the characteristics of popular deep learning GitHub repositories[C]//Proceedings of the 2023 International Conference on Intelligent Computing and Next Generation Networks (ICNGN). Piscataway: IEEE Press, 2023: 1-6.

CHEN X, ZHANG A Y. Achieving optimal clustering in Gaussian mixture models with anisotropic covariance structures[C]//Proceedings of the 38th International Conference on Neural Information Processing Systems (NeurIPS). New York: Curran Associates, 2024: 113698-113741.

ZHOU Z L, LI C, VAN MIEGHEM P, et al. Predicting higher-order dynamics with unknown hypergraph topology[J]. IEEE Transactions on Circuits and Systems I: Regular Papers, 2025, 72(4): 1693-1706.

GÜNEYI E T, YALDıZ B, CANBOLAT A, et al. Learning graph ARMA processes from time-vertex spectra[J]. IEEE Transactions on Signal Processing, 2024, 72: 47-56.

HYNDMAN R J, ATHANASOPOULOS G. Forecasting: principles and practice[M]. Melbourne: OTexts, 2018.

WANG B, YANG A. A consolidated cross-validation algorithm for support vector machines via data reduction[C]//Proceedings of the 36th International Conference on Neural Information Processing Systems (NeurIPS). New York: Curran Associates, 2022, 35: 394-405.

RHODES J S, CUTLER A, MOON K R. Geometry- and accuracy-preserving random forest proximities[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2023, 45(9): 10947-10959.

YAN S, LIANG Y, ZHENG L, et al. Explainable gated Bayesian recurrent neural network for non-Markov state estimation[J]. IEEE Transactions on Signal Processing, 2024, 72: 4302-4317.

HOCHREITER S, SCHMIDHUBER J. Long short-term memory[J]. Neural Computation, 1997, 9(8): 1735-1780.

BENIDIS K, RANGAPURAM S S, FLUNKERT V, et al. Deep learning for time series forecasting: tutorial and literature survey[J]. ACM Computing Surveys, 2022, 55(6): 1-36.

WANG W, LI C, QU B, et al. Predicting epidemic threshold in complex networks by graph neural network[J]. Chaos, 2024, 34(6): 063129.

VASWANI A, SHAZEER N, PARMAR N, et al. Attention is all you need[C]//Proceedings of the 31st International Conference on Neural Information Processing Systems (NeurIPS). New York: Curran Associates, 2017: 5998-6008.

姚云, 胡振虓, 邓涛, 等. 基于自适应池化注意力Transformer的唇语识别方法[J]. 智能科学与技术学报, 2025, 7(2): 211-220.

YAO Y, HU Z X, DENG T, et al. A lip reading method based on adaptive pooling attention Transformer[J]. Chinese Journal of Intelligent Science and Technology, 2025, 7(2): 211-220.

杨波, 于茜, 张伟, 等. GitHub开源软件开发过程中影响因素的相关性分析[J]. 软件学报, 2017, 28(6): 1330-1342.

YANG B, YU Q, ZHANG W, et al. Influence factors correlation analysis in git hub open source software development process[J]. Journal of Software, 2017, 28(6): 1330-1342.

王明宇, 宫庆媛, 瞿晶晶, 等. 基于机器学习的GitHub企业影响力分析与预测[J]. 智能科学与技术学报, 2023, 5(3): 330-342.

WANG M Y, GONG Q Y, QU J J, et al. Analysis and prediction of GitHub company influence based on machine learning[J]. Chinese Journal of Intelligent Science and Technology, 2023, 5(3): 330-342.

SÁNCHEZ R J M, CIRIA G G, OLIVERO M Á, et al. Feature-based characterization of GitHub OSS projects for performance assessment[C]//Proceedings of the 29th ACM International Systems and Software Product Line Conference-Volume A. New York: ACM Press, 2025: 94-104.

HAN J W, KAMBER M, PEI J. Data mining: concepts and techniques[M]. 3rd ed. Amsterdam: Elsevier, 2012.

MURTAGH F, LEGENDRE P. Ward's hierarchical agglomerative clustering method: which algorithms implement ward's criterion?[J]. Journal of Classification, 2014, 31(3): 274-295.

KAUFMAN L, ROUSSEEUW P J. Finding groups in data: an introduction to cluster analysis[M]. New York: John Wiley & Sons, 1990.

GIUDICE E, KUIPERS J, MOFFA G. A Bayesian take on Gaussian process networks[C]//Proceedings of the 37th International Conference on Neural Information Processing Systems (NeurIPS). New York: Curran Associates, 2023: 56602-56614.

BISHOP C M. Pattern recognition and machine learning[M]. New York: Springer, 2006.

CHEN T Q, GUESTRIN C. XGBoost: a scalable tree boosting system[C]//Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. New York: ACM Press, 2016: 785-794.

ZENG A L, CHEN M X, ZHANG L, et al. Are transformers effective for time series forecasting?[C]//Proceedings of the AAAI Conference on Artificial Intelligence. Palo Alto: AAAI Press, 2023, 37(9): 11121-11128.

DAS A, KONG W, LEACH A, et al. Long-term forecasting with TiDE: time-series dense encoder[J]. Transactions on Machine Learning Research, 2023: 1-27.

浏览量

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

大模型驱动的社交智能体谣言易感性与干预策略研究

RadiFlow：AI Agents重构放射科工作流程

融合知识图谱和大语言模型的罗氏沼虾养殖问答系统的设计与实现

面向加密货币反洗钱的智能监管研究：模型、方法与应用

Cerberus：基于深度学习的跨站社交网络机器人检测系统