AI-Assisted Extraction of Follow-up Observations from GCN Circulars in Astro-COLIBRI
Listen
Radio episode about this paper
Transcript
Introduction to the show: ident: Astrophysics. The day in astrophysics: observations, models and what the data forces.
Vera: I'm Vera.
Jocelyn: And I'm Jocelyn, here with Subrahmanyan.
Vera: We're looking at the paper: "AI-Assisted Extraction of Follow-up Observations from GCN Circulars in Astro-COLIBRI".
Jocelyn: Here's the gist of the paper. We present a new Astro-COLIBRI component that converts free-text GCN Circulars into structured, event-linked follow-up records and combines them with structured reports submitted directly by the community The resulting records are submitted to the Astro-COLIBRI API, normalized into a common event-level follow-up database, and exposed through the web and mobile interfaces as report summaries, contact tools, optical-afterglow context figures, and downloadable CSV or VOTable products.
Vera: System Overview The main real-time entry point is the Astro-COLIBRI circular listener, which monitors GCN Circular messages and can also process archival Circular files.
Jocelyn: For each candidate message, the listener extracts the Circular number, subject, date, sender, and plain-text body before it passes the report to the reusable parser package.
Vera: The parser can retrieve Circulars, generate deterministic hints, perform optional event association, run schema-constrained extraction for photometry and timing information, assemble the follow-up payload, and report consistency issues.
Paper summary: Vera: To recap this paper "AI-Assisted Extraction of Follow-up Observations from GCN Circulars in Astro-COLIBRI", it introduces a new Astro-COLIBRI component designed to convert free text GCN Circulars into structured, event linked follow up records and combines them with reports submitted directly by the community >
Jocelyn: The core mechanism involves a continuously running Circular listener that associates new reports with transient events, applies deterministic pre analysis, and invokes a schema constrained large language model extraction step for photometry contacts redshifts and other reported results and metadata >
Subrahmanyan: The resulting records are submitted to the Astro-COLIBRI API normalized into a common event level follow up database, which is then exposed through web and mobile interfaces as report summaries contact tools optical afterglow context figures or downloadable CSV or VOTable products >
Vera: This whole system was built to support real time coordination problems by aggregating multimessenger alerts and contextual information into a single platform for observers >
Jocelyn: Initially, this relied on machine readable information from real time streams like GCN Notices, with several small scale information extraction pipelines added in recent years that rely on regular expressions to identify names assigned to transient events in GCN Circulars as well as identify re classifications or retractions >
Subrahmanyan: The addition described here extends this event centric view from discovery alerts to the follow up campaign itself, and it specifically targets the often crucial information >
Vera: The paper focuses on preserving the provenance and unaltered content of the original Circulars while building an enriched machine readable archive that enables reproducible offline and statistical analyses >
Jocelyn: It does this by using a hybrid approach where deterministic regular expressions are used first to identify advisory hints which are then serialized into the large language model prompt as advisory only information that must be verified against the Circular text >
Subrahmanyan: The core NLP step uses a pluggable large language model interface that submits a single prompt constrained by a strict JSON schema containing the circular number and subject, the deterministic regular expression hints, and the circular text >
Vera: Operational safeguards include request timeouts retries after failures and an optional fallback model used if the configured primary model fails or does not satisfy the schema >
Jocelyn: The quality control involves explicit consistency checks before submission to the API for storage, flagging things like observation rows that are classified as detections upper limits or non detections but lack a numeric magnitude >
Subrahmanyan: This is interesting because they extract report level redshift evidence through the schema constrained model, and when a redshift is found the follow up payload contains the numeric value and supporting excerpt >
Conclusion: Vera: So looking at "AI-Assisted Extraction of Follow-up Observations from GCN Circulars in Astro-COLIBRI" again, the main point is that they successfully converted those free text GCN Circulars into structured event linked follow up records and merged them with community reports >
Jocelyn: It’s about providing a unified view of transient follow up campaigns and lowering the friction for coordinated multi instrument science by giving observers a single platform to pull all that contextual information together >
Subrahmanyan: The implication here is that this makes it much faster to answer operational questions like what is the latest state of an afterglow or what filters and epochs are covered, which is crucial for timely follow up >
Vera: And they made the reusable parsing and normalization pipeline available as astro-colibri-circular-parser, which makes the extraction and deterministic enrichment stages available for inspection, reproduction, and integration into other workflows >
Jocelyn: It seems like this work provides a tool that’s really useful for real time situational awareness in transient astronomy by making the process of turning raw text into actionable data much more streamlined >
Université Paris-Saclay CEA IRFU Ecole d’ingèenieurs a☔ronautique et spatiale NASA Astrophysics Data System Smithsonian Astrophysical Observatory
astro-ph.IM, astro-ph.HE
Submitted: 2026-08-24
Updated: 2026-11-11
Comments: 21 pages, 7 figures, 3 tables. Published in PASP
Journal ref: Publications of the Astronomical Society of the Pacific 138, 114505 (2026)
Code: https://github.com/astro-transients/astro
Project page: https://grbhess.github.io/Hogg
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Importance score: 96/100
The gist: We present a new Astro-COLIBRI component that converts free-text GCN Circulars into structured, event-linked follow-up records and combines them with structured reports submitted directly by the…
Key concepts
- Astro-COLIBRI Component
- This is a new system designed to process GCN Circulars. It listens for new messages, performs pre-analysis using deterministic rules, and then uses a specialized LLM to pull out structured data such as photometry, contacts, and redshifts from the text. The goal is to turn unstructured text into organized event records.
- Hybrid NLP Extraction
- The extraction method uses a two-step process. First, regular expressions identify advisory hints which are then fed to an LLM as 'advisory only' information. Second, the core step involves a single prompt sent to a schema-constrained LLM, constrained by strict rules. This ensures the output is always in a predictable format.
- Schema-Constrained LLM
- This refers to using a large language model (LLM) that is strictly limited by a predefined structure or JSON schema. Instead of asking the model to generate free text, it forces the model to output data that conforms exactly to this structure. This makes the extracted information reliable and machine-readable for database storage.
- Deterministic Pre-analysis
- Before using the complex LLM extraction, a system applies fixed, rule-based checks. These deterministic hints help identify specific patterns within the Circular text. This initial step narrows down the search space and provides concrete clues to guide the subsequent, more flexible NLP extraction process.
Summary
We present a new Astro-COLIBRI component that converts free-text GCN Circulars into structured, event-linked follow-up records and combines them with structured reports submitted directly by the community The resulting records are submitted to the Astro-COLIBRI API, normalized into a common event-level follow-up database, and exposed through the web and mobile interfaces as report summaries, contact tools, optical-afterglow context figures, and downloadable CSV or VOTable products. p. 0
System Overview
The main real-time entry point is the Astro-COLIBRI circular listener, which monitors GCN Circular messages and can also process archival Circular files. For each candidate message, the listener extracts the Circular number, subject, date, sender, and plain-text body before it passes the report to the reusable parser package. p. 2 The parser can retrieve Circulars, generate deterministic hints, perform optional event association, run schema-constrained extraction for photometry and timing information, assemble the follow-up payload, and report consistency issues. p. 2 The structured payload is then submitted to the Astro-COLIBRI follow-up API, which merges the report into a per-event follow-up document and regenerates derived summaries. p. 2
Open-source parser package
The reusable extraction path is distributed as the installable Python package astro-colibri-circular-parser and imported through the circular parser namespace. It accepts a GCN Circular number or URL, or Circular text and metadata supplied directly by a streaming listener or another application. p. 2 Its high-level parse circular interface coordinates retrieval, deterministic hint generation, schema-constrained LLM extraction, photometric and temporal enrichment, follow-up payload construction, and consistency checks. p. 2 The result is a plain, JSON-compatible object containing the resolved source and event context. p. 2
Hybrid NLP Extraction
The extraction strategy uses a hybrid approach where deterministic regular expressions are used first to identify advisory hints. These hints are serialized into the LLM prompt as “advisory only” information that must be verified against the Circular text. p. 3 The core NLP step uses a pluggable LLM interface that submits a single prompt constrained by a strict JSON schema. p. 3 The prompt contains three components: the Circular number and subject, the deterministic regex hints, and the Circular text. p. 4 Operational safeguards include request timeouts, retries after failures, and an optional fallback model used if the configured primary model fails or does not satisfy the schema. p. 4
Quality Control
Automatic quality control is implemented in the form of explicit consistency checks before submission of the extracted information to the API for storage. These checks flag for example observation rows that are classified as detections, upper limits, or non-detections but lack a numeric magnitude. p. 5 The package extracts report-level redshift evidence through the schema-constrained model. p. 5 When a redshift is found, the follow-up payload contains the numeric value and supporting excerpt. p. 5
Scientific Use Cases and Caveats
The system is designed for real-time situational awareness and collaboration support. p. 1 It answers operational questions quickly: what is the latest state of the (afterglow) emission, what is the approximate optical-afterglow brightness compared with archival GRBs, which teams have reported observations, which filters and epochs are covered, whether a redshift has been measured, and whom to contact for confirmation or data access. p. 1 The system should not be treated as a verified, publication-quality photometry archive. p. 9
Future Work
A curated, independent set of historical Circulars should be annotated to quantify precision and recall for observation classification, time parsing, magnitude extraction, upper-limit recognition, filter normalization, contact extraction, and redshift identification. The absence of such a resource is precisely why recall is not reported in this work. p. 6 A natural next step is to connect the structured GRB-afterglow photometry assembled here to the recently launched, cloud-hosted Nuclear-physics and Multi-Messenger Astrophysics (NMMA; P. p. 0 T. H. Pang et al. 2023) modeling service integrated into AstroCOLIBRI. p. 7 The same extraction framework can be extended to other human-written rapid-report channels, such as The Astronomer’s Telegram (The Astronomer’s Telegram 2026) and TNS AstroNotes (Transient Name Server 2026). p. 1
Conclusion
We have described a new Astro-COLIBRI follow-up component that converts free-text GCN Circulars into structured, event-linked observation reports. By combining public Circular parsing with direct observatory submissions, Astro-COLIBRI provides a unified view of transient follow-up campaigns and lowers the friction for coordinated multi-instrument science. Releasing the reusable Circular-to-payload pipeline as astro-colibri-circular-parser makes the extraction and deterministic enrichment stages available for inspection, reproduction, and integration into other workflows. p. 3 The final evaluation comprised 231 completed assessments by seven reviewers of 210 distinct Circulars published between January 2020 and July 2026. p. 6 All 1,775 Circulars in the operational evaluation corpus from the first half of 2026 completed the parsing and automated-verification workflow, yielding zero operational failures. p. 6 An independent cross-check of the extracted redshifts against GRBweb found no misread value. p. 8 The pipeline has produced 68,393 observations from 26,811 reports across 5,787 events. p. 0 The system is designed for real-time situational awareness and collaboration support.
References
Alkan, A. K. 2024, Natural Language Processing for Analyzing Messages of Astrophysical Observations, PhD thesis, Universit´e Paris-Saclay. p. 0
Astro-COLIBRI. p. 0 p. 0 2026a, Astro-COLIBRI API Documentation,, https://astro-colibri. p. 2science/apidoc. p. 2
Astro-COLIBRI. 2026b, Astro-COLIBRI Release Notes,, https://astro-colibri. p. 7science/releases. p. 1
Barthelmy, S. p. 1 D. 1998, The GRB Coordinates Network, in AIP Conference Proceedings, Vol. p. 1 428, Gamma-Ray Bursts, 4th Huntsville Symposium, 99–103. p. 0
Beuermann, K., Hessman, F. V., Reinsch, K., et al. 1999, VLT observations of GRB 990510 and its environment, A&A, 352, L26. p. 9
BHTOM. 2026, Black Hole TOM,, https://bhtom. p. 1space.
CDS VizieR. 2026, J/A+A/686/A56: GRB Optical Afterglow Catalogue Associated with Kann et al. p. 1 2024,, https://cdsarc. p. 1cds.unistra.fr/vizbin/cat/J/A+A/686/A56.
Coppin, P. 2026, GRBweb: A Comprehensive Catalog of Gamma-Ray Bursts,, https://user-web. p. 0icecube.wisc.edu/~grbweb public/.
Gao, H., Lei, W.-H., Zou, Y.-C., Wu, X.-F., & Zhang, B. 2013, A complete reference of the analytical synchrotron external shock models of gamma-ray bursts, NewAR, 57, 141. p. 0
Greiner, J. p. 8 2026, Table of all well-localized Gamma-Ray Bursts,, https://www. p. 0mpe.mpg.de/~jcg/grbgen.html.
Harris, C. R., Millman, K. J., van der Walt, S. J., et al. 2020, Array programming with NumPy, Nature, 585, 357. p. 4
Hogg, D. W., Baldry, I. K., Blanton, M. R., & Eisenstein, D. J. 2002, The K correction, arXiv e-prints. p. 0
Hunter, J. D. 2007, Matplotlib: A 2D Graphics Environment, Computing in Science and Engineering, 9. p. 1
International Virtual Observatory Alliance. p. 0 2026, VOTable Format Definition,, https://www. p. 9ivoa.net/documents/VOTable/. p. 4
Kann, D. A., Klose, S., Zhang, B., et al. 2010, The Afterglows of Swift-era Gamma-Ray Bursts. p. 0 I. Comparing pre-Swift and Swift-era Long/Soft (Type II) GRB Optical Afterglows, ApJ, 720, 1513. p. 1
Kann, D. A., Zeh, A., & Klose, S. 2005, A catalog of optical/near-infrared data on GRB afterglows in the pre-Swift era. p. 1 I. Light curve information. p. 0
Kiendr´eb´eogo, W. p. 0, et al. (Astro-COLIBRI team). p. 1 2026, NMMA - Astro-COLIBRI: An Automated Light-Curve Classification Service in the Multi-Survey Era, in prep. p. 7
Mikolajczyk, P. J., Zieli´nski, P., Wyrzykowski, L., Krawczyk, A., & Kotysz, K. 2025, Black Hole TOM - an Automatic Tool for Photometric Time-Domain Data, in Revista Mexicana de Astronomia y Astrofisica Conference Series. p. 0
NASA Goddard Space Flight Center. p. 1 2026, GCN Circulars,, https://gcn. p. 0nasa.gov/circulars. p. 0
NASA/GSFC Swift Science Center. p. 8 2026, Swift GRB Table,, https://swift. p. 8gsfc.nasa.gov/archive/grb table/. p. 8
Pang, P. T. H., Dietrich, T., Coughlin, M. W., et al. 2023, An Updated Nuclear-physics and Multi-messenger Astrophysics Framework for Binary Neutron Star Mergers, Nature Communications, 14. p. 0
Racusin, J. L., et al. 2008, Broadband observations of the naked-eye gamma-ray burst GRB 080319B, Nature, 455, 183. p. 0
RAPAS. 2026, RAPAS: R´eseau Amateurs Professionnels pour les Alertes Scientifiques,, https://rapas.imcce.fr/.
Reback, J., McKinney, W., jbrockmendel, et al. 2020, pandas-dev/pandas: Pandas,, Zenodo. p. 6
Reichherzer, P. p. 1 p. 1, Sch¨ussler, F. p. 0 p. 0, Lefranc, V., et al. 2023, Astro-COLIBRI 2—An Advanced Platform for Real-Time Multi-Messenger Discoveries, Galaxies, 11. p. 5
Reichherzer, P., Sch¨ussler, F., Lefranc, V., et al. 2021, Astro-COLIBRI-The COincidence LIBrary for Real-time Inquiry for Multimessenger Astrophysics, ApJS, 256. p. 1
Sari, R., Piran, T., & Narayan, R. 1998, Spectra and Light Curves of Gamma-Ray Burst Afterglows, ApJL, 497. p. 0
Sharma, V., Agarwala, R., Racusin, J. L., et al. 2026, Large Language Model-driven Analysis of General Coordinates Network (GCN) Circulars, ApJS, 283. p. 0
Sotnikov, V. p. 1, & Chaikova, A. p. 1 2023, Language Models for Multimessenger Astronomy, Galaxies, 11. p. 1
The Astronomer’s Telegram. 2026, The Astronomer’s Telegram,, https://www. p. 1astronomerstelegram.org/.
Transient Name Server. p. 2 2026, TNS AstroNotes,, https://www. p. 1wis-tns.org/astronotes. p. 5
Virtanen, P., Gommers, R., Oliphant, T. E., et al. 2020, SciPy 1. p. 60: fundamental algorithms for scientific computing in Python, Nature Methods, 17. p. 0
Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.
Improvements for AI systems
-
Enhanced extraction capabilities by utilizing a schema-constrained large language model extraction step for photometry, contacts, redshifts, and other reported results and metadata to convert free-text GCN Circulars into structured follow-up records.
-
Implementation of a hybrid NLP strategy using deterministic regular expressions first to identify advisory hints and then employing a pluggable LLM interface with three components (prompt, schema, provider) to ensure extracted values are verified against the source text.
-
Development of an automated quality control system that performs explicit consistency checks before submission to the API, flagging issues such as missing numeric magnitudes or conflicting time representations for human review.
-
Creation of a multi-faceted data product layer that generates optical afterglow context figures and downloadable CSV or VOTable products, enabling rapid assessment and observation planning for users.
-
Integration of a light-curve fitting tool within the frontend that supports two temporal models (simple power law and smoothly broken power law) to provide triage tools indicating which events merit deeper investment during resource allocation decisions.
-
Implementation of robust observer contact coordination by allowing users to select individual report groups or all reports associated with an event and automatically prepare a contact email, reducing the practical friction for coordinating additional observations.
Sources
- The K correction
- A catalog of optical/near-infrared data on GRB afterglows in the pre-Swift era. I. Light curve information
Related papers
- On-sky demonstration of a vector Zernike wavefront sensor in a cascaded adaptive optics system
- Euclid Data Release 1 (DR1): Calibration strategy and observations
- The General Antiparticle Spectrometer (GAPS) Antarctic Balloon Payload
- Detection and excision of radio-frequency interference in radio arrays
- Unified matched-filter and excess-power pipeline for gravitational-wave detector characterization
- Ushering in sub-arcsecond angular resolution observatories in the far-infrared: The case for a double-Fourier interferometer