A Network-driven Framework for Public Event Forecasting via Dynamic Interaction Network Evolution

Jie Wei; Yue Liu; Xiaochuan Tang; Biao Cai; Xiangtao Li; Yanmei Hu

cs.AIAug 16, 2026

A Network-driven Framework for Public Event Forecasting via Dynamic Interaction Network Evolution

Authors: Jie Wei, Yue Liu, Xiaochuan Tang, Biao Cai, Xiangtao Li, Yanmei Hu

Organizations: Chengdu University of Technology, Chengdu, 610000, China. · Chengdu Neusoft University, Chengdu, 610000, China. · Jilin University, Changchun, 130000, China.

Abstract

Effective public event forecasting is essential for intelligent service systems, enabling proactive risk management, adaptive resource allocation, and timely decision-making. In many real-world scenarios, the evolution of public events is driven by dynamic interactions among participants. Motivated by this observation, this paper proposes auto-ibDLM, a network-driven deep learning framework that represents events as dynamic interaction networks and predicts public event evolution through participant growth forecasting. The proposed framework adopts a hybrid representation learning strategy that first represents network evolution using network science-informed structural metrics and subsequently transforms the resulting structural feature vectors into compact and robust latent representations through an auto-learning layer. A GRU-based temporal forecasting module is then employed to capture temporal dependencies and predict future participant growth. Extensive experiments on 13 real-world public event datasets and two publicly available dynamic network datasets demonstrate that auto-ibDLM consistently outperforms representative state-of-the-art methods in both forecasting accuracy and generalization capability, achieving over 97% accuracy in public event forecasting. Comprehensive experimental analyses further validate the effectiveness of the proposed hybrid representation learning strategy and demonstrate its representation-level interpretability. These results indicate that auto-ibDLM provides an effective and practical solution for intelligent public event forecasting.

Figures & tables

Figure 1: The architecture of auto-ibDLM. The proposed framework represents an event as a dynamic interaction network and formulates public event forecasting as forecasting future participant growth, represented by node increments in the evolving interaction network. The framework consists of four stages: problem modeling, structural representation of the dynamic interaction network, auto-learning representation, and GRU-based temporal forecasting.

Figure 2: An example of the dynamic interaction network.

Metric	Formula
D	$D(u)=\left\|N_{u}\right\|$
EXTD	$EXTD(u)=d_{u}+\sum_{v\in N(u)}d_{v}$
ACCD	$ACCD(u)=d_{u}+\sum_{v\in N(u)}(d_{v}+\sum_{w\in N(v)}d_{w})$
NM	$NM(u)=\left\|\left\{(v,w)\in\mathcal{E}\mid v,w\in N^{+}(u)\right\}\right\|$
CE	$CE(u)=\frac{\left\|\left\{\left(v,w\right)\in\mathcal{E}\mid v\in N^{+}\left(u\right),w\in\mathcal{V}-N^{+}\left(u\right)\right\}\right\|}{min\left(vol\left(N^{+}\left(u\right)\right)\right),vol\left(\mathcal{V}-N^{+}\left(u\right)\right)}$
DE	$DE(u)=\frac{\left\|\left\{(v,w)\in\mathcal{E}\mid v,w\in N^{+}(u)\right\}\right\|}{\left\|\left\{\left(v,w\right)\mid v,w\in N^{+}\left(u\right)\right\}\right\|}$

Table 1: Node-level structural metrics (

N(u)

is the set of

u

’s neighbors, and

N^{+}(u)

is the combination of

N(u)

and

u

). Details of these metrics are referred to Hu et al. (2022) .

Figure 3: Temporal variations of the out-DE structural metric (Mean and Std) and the corresponding participant and interaction growth. The horizontal axis represents time steps. The left vertical axis represents participant growth (node increment) and interaction growth (edge increment), while the right vertical axis represents the temporal variations of the Mean and Std of the out-DE structural metric.

Figure 4: Temporal variations of the in-LCC structural metric (Mean and Std) and the corresponding participant and interaction growth. The horizontal axis represents time steps. The left vertical axis represents participant growth (node increment) and interaction growth (edge increment), while the right vertical axis represents the temporal variations of the Mean and Std of the in-LCC structural metric.

Events ID	Events Topic	Time interval	#Node	#Edge
E1	This round of the epidemic has spread to 19 provinces	4 days	3659	2951
E2	The whole city of Xuzhou suspends offline teaching	4 days	5749	3636
E3	A special case of a patient infecting thirteen patients appeared in Chengdu	7 days	5126	7026
E4	Advise Beijing citizens not to go out tomorrow unless necessary	4 days	9996	8149
E5	Xuzhou city-wide subway suspension	3 days	7355	5214
E6	Shanghai epidemic	5 days	5288	6167

Table 2: Summary of public events in Weibo benchmark dataset

Figure 5: Comparison of participant growth forecasting results on the Weibo benchmark. EvolveGCN, WinGNN, SiGNN, and ARIMA do not utilize community evolution behaviors and therefore produce identical results under LPA and LMC-APR. The horizontal axis represents the test sample index, and the vertical axis represents the predicted participant growth (i.e., node increment).

Figure 6: Contribution scores of structural metrics derived from the auto-learning weight matrix (quantified by the mean absolute weight of each row).

Removed structural metric	MSE	Relative increase
None	0.000229	–
Std of in-COREDPA	0.000311	+35.8%
Std of out-EXTD	0.000425	+85.6%
Std of out-COREDP	0.000387	+69.0%
Top-3 structural metrics	0.000528	+130.6%

Table 3: Validation of the contribution analysis by removing the top-ranked structural metrics, A lower MSE value indicates superior predictive performance.

Figure 7: Contribution scores of structural metrics derived from the auto-learning weight matrix (quantified by the mean squared sum per row).

Baseline	200		300		400		500		600
	LPA	LMC-APR	LPA	LMC-APR	LPA	LMC-APR	LPA	LMC-APR	LPA	LMC-APR
auto-ibDLM	97.3%	98.7%	100%	100%	100%	100%	100%	100%	97.4%	100%
Transformer	94.7%	92.1%	96.1%	97.4%	97.4%	97.4%	97.4%	96.1%	97.4%	97.4%
LSTM	94.7%	94.7%	96.1%	96.1%	96.1%	97.4%	96.1%	97.4%	97.4%	97.4%
CNN-LSTM	92.1%	93.4%	96.1%	94.7%	97.4%	96.1%	97.4%	97.4%	97.4%	97.4%
VAR	28.9%	28.9%	27.6%	27.6%	27.6%	27.6%	27.6%	27.6%	27.6%	27.6%

Table 4: Accuracy (%) of different models for public event identification under different participant growth thresholds.

Baseline	200		300		400		500		600
	LPA	LMC-APR	LPA	LMC-APR	LPA	LMC-APR	LPA	LMC-APR	LPA	LMC-APR
auto-ibDLM	100%	100%	100%	100%	100%	100%	100%	100%	0	100%
Transformer	33.3%	33.3%	50%	50%	50%	50%	50%	0	0	0
LSTM	0	33.3%	0	0	0	0	0	0	0	0
CNN-LSTM	0	0	0	0	0	0	0	0	0	0
VAR	100%	100%	100%	100%	100%	100%	100%	100%	100%	100%

Table 5: Recall (%) of different models for public event identification under different participant growth thresholds.

Datasets	Weibo		Superuser		Bitcoin
Models	LPA	LMC-APR	LPA	LMC-APR	LPA	LMC-APR
auto-ibDLM	0.000126	0.000071	0.000094	0.000085	0.000627	0.000400
Transformer	0.000521	0.000521	0.000154	0.000100	0.004592	0.003486
LSTM	0.000515	0.000442	0.000114	0.000111	0.003944	0.002000
CNN-LSTM	0.000547	0.000651	0.000259	0.000198	0.006087	0.006127
VAR	0.31161	0.31235	0.007623	0.007871	0.033059	0.038454

Table 6: MSEs of different models across all datasets. The best results are highlighted in bold, and “-” indicates that the corresponding model fails to produce a valid prediction.

Datasets	Weibo		Superuser		Bitcoin
Models	LPA	LMC-APR	LPA	LMC-APR	LPA	LMC-APR
auto-ibDLM	0.000126	0.000071	0.000094	0.000084	0.000627	0.000400
ibDLM	0.000325	0.000486	0.000107	0.000095	0.003973	0.001281
auto-ibDLM ${}_{-loss_{f}}$	0.000423	0.000373	0.000114	0.000101	0.001496	0.001105
manual-ibDLM	0.000249	0.000225	0.000118	0.000095	0.002509	0.000764
auto-ibDLM c+l	0.000287	0.000210	0.000094	0.000088	0.000652	0.000506

Table 7: MSEs of auto-ibDLM and its variants across all datasets. The best results are highlighted in bold.

Explore similar work

SURGE: An Event-Centric Social Media Sentiment Time Series Benchmark with Interaction Structure
May 20, 2026Chen Su, Pengsen Cheng, Yuanhe Tian +1Social Media Analysis Multivariate Time Series Forecasting
University of Science and Technology of China · Sichuan University · Zhongguancun Academy
GTIN: A Unified Framework for Joint Event and Time Prediction in Temporal Graphs
Jul 26, 2026Mohammad Ostadmohammadi, Sepehr Kazemi, Hamid R. RabieeEvent Time Prediction Dynamic Graph Learning
Department of Computer Engineering, Sharif University of Technology, Azadi Ave., Tehran, Iran · Department of Electrical Engineering, Sharif University of Technology, Azadi Ave., Tehran, Iran · Sharif University of Technology, Azadi Ave., Tehran, Iran
FDN: Interpretable Spatiotemporal Forecasting with Future Decomposition Networks
Jun 23, 2026Nicholas Majeske, Ariful AzadExplainable Time Series Forecasting Time Series Forecasting
Department of Intelligent Systems Engineering, Indiana University Bloomington, USA · Department of Computer Science and Engineering, Texas A&M University, USA

May 20, 2026J/K move · Enter open · S save

SURGE: An Event-Centric Social Media Sentiment Time Series Benchmark with Interaction Structure

Chen Su, Pengsen Cheng, Yuanhe Tian, Yan Song

University of Science and Technology of China · Sichuan University · Zhongguancun Academy

Social Media Analysis Multivariate Time Series Forecasting Time Series Forecasting Forecasting Benchmarks

Open paper AskPDF arXiv

Public events on social media generate large volumes of discussion whose collective dynamics carry direct value for opinion forecasting and crisis response. Capturing how these dynamics evolve across an event's lifecycle requires organizing fragmented posts into event-level time series. Existing datasets cover only a small number of events within a single category, and typically discard the interaction structure between posts when constructing time series, which restricts both transfer across event types and controlled study of how interactions shape the resulting collective dynamics. We present SURGE, a multi-event social media benchmark that pairs event-level time series with aligned text and interaction structure linking posts within an event. SURGE is built through an automated pipeline that produces calendar-aligned time series at three temporal granularities, covering 67 events and more than 800K posts across five event categories. Each time bin is paired with flat and structured textual views derived from the same selected posts, enabling controlled evaluation of whether social interaction structure affects forecasting behavior. On top of SURGE we define benchmark protocols for numerical-only forecasting, text-augmented forecasting, high-interaction evaluation, and leave-one-category-out generalization. Experiments with representative time-series and multimodal forecasting models reveal three properties of the benchmark: a strong local-persistence regime in which naive baselines remain hard to beat under absolute error, limited transfer of existing text-augmented forecasters to event-driven social-media data, and increased difficulty on reply-dense periods that aggregate metrics tend to obscure. We further include a lightweight structure-aware probe as a reference implementation, illustrating how SURGE can support interaction-aware forecasting research.

Abstract

Explore similar work

SURGE: An Event-Centric Social Media Sentiment Time Series Benchmark with Interaction Structure

GTIN: A Unified Framework for Joint Event and Time Prediction in Temporal Graphs

FDN: Interpretable Spatiotemporal Forecasting with Future Decomposition Networks