MissingNarrative-AEDL: Reliability-Aware Causal Inference Under Missing and Conflicting Event Evidence

Authors

  • Marie Dubois Department of Industrial and Systems Engineering, University at Buffalo, Buffalo, NY, USA Author
  • Amir Qureshi School of Electrical Engineering and Computer Science, Oregon State University, Corvallis, OR, USA Author
  • Helena Braun School of Electrical Engineering and Computer Science, Oregon State University, Corvallis, OR, USA Author

DOI:

https://doi.org/10.66525/ijictt107

Keywords:

causal inference, missing evidence, conflicting narratives, proxy SVAR, neuro-symbolic AI, reliability learning, non-stationary time series.

Abstract

Event-driven causal inference increasingly depends on narrative evidence such as news, incident reports, policy statements, and expert annotations. Existing event-driven labeling pipelines have shown that semantic narratives can strengthen causal identification in non-stationary time series, but practical deployments rarely observe complete and internally consistent evidence. Reports may be missing, delayed, duplicated, low-quality, or mutually contradictory. This paper proposes MissingNarrative-AEDL, a reliability-aware extension of adaptive event-driven labeling for causal inference under missing and conflicting event evidence. The method treats each evidence source as an online reliability process, fuses available event claims through a conflict-penalized consensus operator, and sends only verified instruments to a proxy structural vector autoregression. A small expert-audited calibration window is used to initialize source reliability; subsequent operation is fully online and uses only market-response feedback and cross-source coherence. To avoid unsupported empirical claims, experiments are performed on a fully reproducible non-stationary causal benchmark generated by the attached code, with latent supply, demand, and uncertainty shocks, six heterogeneous evidence sources, endogenous missingness, and high-confidence conflicting reports. The proposed method achieves the highest macro F1 score of 0.920 and supply-shock F1 score of 0.929, compared with 0.870 for a reliability-unaware AEDL variant and 0.755 for majority narrative voting. Across a 24-scenario stress test, MissingNarrative-AEDL maintains the highest average macro F1 of 0.889 while preserving strong proxy relevance with an average first-stage F statistic of 85.3. The results indicate that reliability-aware narrative instruments can improve causal signal extraction without fabricating missing evidence or assuming perfect source agreement.

Downloads

Download data is not yet available.

References

] A. Vaswani et al., "Attention is all you need," in Advances in Neural Information Processing Systems, 2017.

[2] D. Teng, "TEAS: Token- and Energy-Aware Autoscaling for Cost-Efficient LLM Serving," AI and Data Science Journal, 2025

[3] R. J. A. Little and D. B. Rubin, Statistical Analysis with Missing Data, 3rd ed. Hoboken, NJ, USA: Wiley, 2019.

[4] K. Mohan, J. Pearl, and J. Tian, "Graphical models for inference with missing data," in Advances in Neural Information Processing Systems, 2013.

[5] A. P. Dempster, "Upper and lower probabilities induced by a multivalued mapping," Annals of Mathematical Statistics, vol. 38, no. 2, pp. 325-339, 1967.

[6] G. Shafer, A Mathematical Theory of Evidence. Princeton, NJ, USA: Princeton University Press, 1976.

[7] J. Pearl, Causality: Models, Reasoning, and Inference, 2nd ed. Cambridge, U.K.: Cambridge University Press, 2009.

[8] J. L. Montiel Olea, J. H. Stock, and M. W. Watson, "Inference in structural vector autoregressions identified with an external instrument," Journal of Econometrics, vol. 225, no. 1, pp. 74-87, 2021.

[9] K. Mertens and M. O. Ravn, "The dynamic effects of personal and corporate income tax changes in the United States," American Economic Review, vol. 103, no. 4, pp. 1212-1247, 2013.

[10] L. Kilian, "Not all oil price shocks are alike: Disentangling demand and supply shocks in the crude oil market," American Economic Review, vol. 99, no. 3, pp. 1053-1069, 2009.

[11] C. Baumeister and J. D. Hamilton, "Structural interpretation of vector autoregressions with incomplete identification: Revisiting the role of oil supply and demand shocks," American Economic Review, vol. 109, no. 5, pp. 1873-1910, 2019.

[12] D. Caldara and M. Iacoviello, "Measuring geopolitical risk," American Economic Review, vol. 112, no. 4, pp. 1194-1225, 2022.

[13] R. J. Shiller, "Narrative economics," American Economic Review, vol. 107, no. 4, pp. 967-1004, 2017.

[14] D. M. Blei, A. Y. Ng, and M. I. Jordan, "Latent Dirichlet allocation," Journal of Machine Learning Research, vol. 3, pp. 993-1022, 2003.

[15] N. Reimers and I. Gurevych, "Sentence-BERT: Sentence embeddings using Siamese BERT-networks," in Proc. EMNLP-IJCNLP, 2019, pp. 3982-3992.

[16] Y. Liang, Y. Jiao, W. Ping, H. Fan, and X. Han, "Adaptive Event-Driven Labeling: A Neuro-Symbolic Multiagent Framework for Causal Inference in Non-Stationary Time Series," IEEE Access, 2026.

[17] J. Devlin, M.-W. Chang, K. Lee, and K. Toutanova, "BERT: Pre-training of deep bidirectional transformers for language understanding," in Proc. NAACL-HLT, 2019, pp. 4171-4186.

[18] J. Wei et al., "Chain-of-thought prompting elicits reasoning in large language models," in Advances in Neural Information Processing Systems, 2022.

[19] Y. Du, S. Li, A. Torralba, J. B. Tenenbaum, and I. Mordatch, "Improving factuality and reasoning in language models through multiagent debate," arXiv:2305.14325, 2023.

[20] N. Shinn, F. Cassano, E. Berman, A. Gopinath, K. Narasimhan, and S. Yao, "Reflexion: Language agents with verbal reinforcement learning," in Advances in Neural Information Processing Systems, 2023.

[21] J. Kaddour, A. Lynch, Q. Liu, M. J. Kusner, and R. Silva, "Causal machine learning: A survey and open problems," arXiv:2206.15475, 2022.

[22] P. Spirtes, C. Glymour, and R. Scheines, Causation, Prediction, and Search, 2nd ed. Cambridge, MA, USA: MIT Press, 2000.

[23] D. M. Chickering, "Optimal structure identification with greedy search," Journal of Machine Learning Research, vol. 3, pp. 507-554, 2002.

[24] J. Peters, D. Janzing, and B. Scholkopf, Elements of Causal Inference: Foundations and Learning Algorithms. Cambridge, MA, USA: MIT Press, 2017.

[25] S. van Buuren and K. Groothuis-Oudshoorn, "mice: Multivariate imputation by chained equations in R," Journal of Statistical Software, vol. 45, no. 3, pp. 1-67, 2011.

[26] J. Yoon, J. Jordon, and M. van der Schaar, "GAIN: Missing data imputation using generative adversarial nets," in Proc. 35th International Conference on Machine Learning, 2018, pp. 5689-5698.

[27] B. Zi, "Cloud-native distributed systems for real-time payment intelligence," AI and Data Science Journal, 2024.

Downloads

Published

2026-07-29