Compare commits

..

1 Commits

Author SHA1 Message Date
codex-bot 28805462ee Add reproducible CIKM 2026 source collection
Metadata integrity checks / metadata (push) Successful in 7m34s
2026-11-18 15:49:13 +08:00
7 changed files with 209 additions and 349 deletions
+20
View File
@@ -47,6 +47,26 @@ jobs:
subprocess.run([python,'-B','-m','pip','install','--no-deps',str(wheel)],check=True) subprocess.run([python,'-B','-m','pip','install','--no-deps',str(wheel)],check=True)
outside = root / 'outside-checkout' outside = root / 'outside-checkout'
outside.mkdir() outside.mkdir()
cikm_snapshots = pathlib.Path('/data/paper-index/validation/cikm2026-source-snapshots-v1')
if cikm_snapshots.is_dir():
cikm_output = root / 'installed-cikm2026'
command = [python, '-I', '-B', '-m', 'paper_atlas', 'collect', 'cikm2026-sources']
subprocess.run([*command, '--snapshots', str(cikm_snapshots), '--out', str(cikm_output)], cwd=outside, check=True)
report = json.loads((cikm_output/'collection-report.json').read_text())
assert {name: entry['records'] for name, entry in report['files'].items()} == {
'programme.jsonl': 1008, 'alternate-programme.jsonl': 1005,
'supplemental-programme.jsonl': 73, 'crossref.jsonl': 1081}
assert len(report['registry_outside_doi_programme']) == 73 and not report['database_written']
programme = [json.loads(line) for line in (cikm_output/'programme.jsonl').read_text().splitlines()]
assert all(row['abstract'] and row['author_display_names'] for row in programme)
assert len({row['doi'] for row in programme}) == 1008
broken = root/'tampered-cikm2026'
broken.mkdir()
(broken/'program.html').write_text('tampered')
rejected = subprocess.run([*command, '--snapshots', str(broken), '--out', str(root/'rejected-cikm2026')], cwd=outside, capture_output=True, text=True)
assert rejected.returncode != 0 and 'Snapshot hash mismatch' in rejected.stderr
assert not (root/'rejected-cikm2026').exists()
print('CIKM 2026 full frozen-source CLI check passed', flush=True)
mm_fixture = json.loads(gzip.decompress((src/'tests/fixtures/mm2026-source.json.gz').read_bytes())) mm_fixture = json.loads(gzip.decompress((src/'tests/fixtures/mm2026-source.json.gz').read_bytes()))
mm_snapshots = root/'mm2026-snapshots' mm_snapshots = root/'mm2026-snapshots'
for name, body in mm_fixture['files'].items(): for name, body in mm_fixture['files'].items():
+2
View File
@@ -137,6 +137,8 @@ ECCV 2026 的[暂定节目采集器](docs/eccv2026-program.md)合并重复节目
`collect sdm2026-proceedings` 核验 53 个 Crossref 论文章节,并单独保留 53 项日程观察及 52 页摘要。作者差异、取消展示和摘要缺失分别保留,不自动拼接来源或绑定作品;见 [SDM 2026 使用与恢复说明](docs/sdm2026-proceedings.md)。 `collect sdm2026-proceedings` 核验 53 个 Crossref 论文章节,并单独保留 53 项日程观察及 52 页摘要。作者差异、取消展示和摘要缺失分别保留,不自动拼接来源或绑定作品;见 [SDM 2026 使用与恢复说明](docs/sdm2026-proceedings.md)。
`collect cikm2026-sources` 核验并分开导出两份官方节目、附属议程和出版登记。见 [CIKM 2026 来源采集与恢复](docs/cikm2026-sources.md)。
`collect icdm2026-program` 离线核验 324 条 ICDM2026 官方线下日程,按明确的研究/应用标签输出 241/83 条;展示长短、缺失摘要及未核实出版身份分别保留,不自动入库。见 [ICDM2026 使用与恢复说明](docs/icdm2026-program.md)。 `collect icdm2026-program` 离线核验 324 条 ICDM2026 官方线下日程,按明确的研究/应用标签输出 241/83 条;展示长短、缺失摘要及未核实出版身份分别保留,不自动入库。见 [ICDM2026 使用与恢复说明](docs/icdm2026-program.md)。
`collect recsys2026-program` 核验九个 track 的 263 项官方录用贡献观察(260 个不同 DOI),保留摘要、原始作者和重复 DOI 的来源版本。见 [RecSys 2026 使用与恢复说明](docs/recsys2026-program.md)。 `collect recsys2026-program` 核验九个 track 的 263 项官方录用贡献观察(260 个不同 DOI),保留摘要、原始作者和重复 DOI 的来源版本。见 [RecSys 2026 使用与恢复说明](docs/recsys2026-program.md)。
+22
View File
@@ -0,0 +1,22 @@
# CIKM 2026 来源采集与恢复
使用固定节目和登记快照复现 CIKM 2026 来源盘点。采集命令生成四份独立 JSONL,输入逐文件校验 SHA-256。
```sh
paper-atlas collect cikm2026-sources \
--snapshots /data/paper-index/validation/cikm2026-source-snapshots-v1 \
--out /data/paper-index/collected/cikm2026-sources-v1
```
输入包括两版官方节目 HTML、Crossref 卷登记、五页逐篇登记及分页回执。服务器验证目录保留输入原文。输出目录必须是新目录。先检查 `collection-report.json`,报告在四份 JSONL 写完后发布。
| 文件 | 内容 |
| --- | --- |
| `programme.jsonl` | 带 DOI 的 1,008 条节目记录,包括标题、作者显示名、摘要和五类节目标签 |
| `alternate-programme.jsonl` | 另一版节目 1,005 条记录及各自的字段 |
| `supplemental-programme.jsonl` | 73 条演讲、教程、工作坊及其他附属活动观察 |
| `crossref.jsonl` | 1,081 条 ACM 卷内登记,每条保留登记字段及分页文件位置 |
节目 DOI 指向 Crossref 的出版登记。确认作者和标题差异以及历史投稿绑定后,再为来源观察准备入库队列。报告另列 73 个未出现在带 DOI 节目区的登记,附属活动仍保留各自的活动类型。公开拒稿与撤回决定仍需独立来源核查。
官方原始来源见[带 DOI 的节目](https://cikm2026.diag.uniroma1.it/cikm-program/program.html)、[另一版节目](https://cikm2026.diag.uniroma1.it/cikm-program/CIKM2026_program.html)和 [Crossref 卷登记](https://api.crossref.org/works/10.1145/3799682)。
@@ -1,339 +0,0 @@
{
"sources": {
"poster.html": {
"url": "https://www3.cs.stonybrook.edu/~icdm2026/PosterwDemo/list.htm",
"final_url": "https://www3.cs.stonybrook.edu/~icdm2026/PosterwDemo/list.htm",
"http_status": 200,
"bytes": 132977,
"sha256": "b02083ca7574a188fe876851a700a12c131c262353db557bd43a8320fad5f747",
"at": "2026-11-18T07:06:11.984184+00:00"
},
"schedule.html": {
"url": "https://icdm2026.neu.edu.cn/DetailedSessions/list.htm",
"final_url": "https://icdm2026.neu.edu.cn/DetailedSessions/list.htm",
"http_status": 200,
"bytes": 213830,
"sha256": "c1294b89bae8c2dfcf652d8a25556e733ef42dc3ac3387070abe08419b961b0f",
"at": "2026-11-18T07:06:12.349258+00:00"
},
"deepsync.html": {
"url": "https://dm.cs.univie.ac.at/news/story/deepsync-paper-accepted-at-icdm-2026",
"final_url": "https://dm.cs.univie.ac.at/news/story/deepsync-paper-accepted-at-icdm-2026",
"http_status": 200,
"bytes": 90333,
"sha256": "e243196943434225acec2841b53c77a2b1459184e7c185e1e1cbb055118fe282",
"at": "2026-11-18T07:06:20.561308+00:00"
}
},
"poster_count": 343,
"poster_groups": {
"long-posters": 177,
"short-posters": 166
},
"oral_count": 324,
"shared_count": 324,
"poster_only": {
"DM1022": {
"group": "short-posters",
"floor": "3F",
"area": "Area C",
"slot": "P2-3C-085",
"session": "Applied Track: Real-world AI in finance, healthcare and industry II",
"title": "Beyond Bidding Agents: A Dual-Role Multi-Agent Framework with Dynamic Reserve Pricing for Auto-Bidding Auctions",
"line": 683
},
"DM1189": {
"group": "long-posters",
"floor": "3F",
"area": "Area A",
"slot": "P1-3A-032",
"session": "Heterogeneous data mining: spatiotemporal and representation learning",
"title": "Traj-Transformer: Diffusion Models with Transformer for GPS Trajectory Generation",
"line": 456
},
"DM1234": {
"group": "long-posters",
"floor": "2F",
"area": "Area A",
"slot": "P1-2A-007",
"session": "Foundations, causality and learning theory",
"title": "Prediction-Powered Corpus Mining: Coverage-Valid Statistics from Scarce Labels",
"line": 354
},
"DM1246": {
"group": "short-posters",
"floor": "3F",
"area": "Area B",
"slot": "P2-3B-074",
"session": "LLM-driven data mining and novel applications II",
"title": "Feature Selection for Trajectory Inference",
"line": 672
},
"DM1345": {
"group": "short-posters",
"floor": "3F",
"area": "Area B",
"slot": "P2-3B-060",
"session": "LLM-driven data mining and heterogeneous learning",
"title": "TRACER: An LLM-based Coarse-to-Fine Framework for Adaptive Network Traffic Classification",
"line": 658
},
"DM1408": {
"group": "long-posters",
"floor": "2F",
"area": "Area B",
"slot": "P1-2B-043",
"session": "Applied Track: Industrial, financial and decision intelligence",
"title": "A Generation-to-Application Framework for Semantic ID Collision in Industrial Generative Recommendation",
"line": 390
},
"DM1644": {
"group": "long-posters",
"floor": "2F",
"area": "Area B",
"slot": "P1-2B-072",
"session": "Recommendation, adaptation and spatiotemporal modeling II",
"title": "LQ-Rec: Late Quantization for Sequential Recommendation",
"line": 419
},
"DM1751": {
"group": "short-posters",
"floor": "3F",
"area": "Area A",
"slot": "P2-3A-014",
"session": "ML/DL and statistical methods",
"title": "MissGNN: Missingness-Guided Graph Neural Network for Feature- and Sample-Aware Imputation",
"line": 612
},
"DM1995": {
"group": "long-posters",
"floor": "2F",
"area": "Area B",
"slot": "P1-2B-053",
"session": "Applied Track: Recommendation, security and operational intelligence II",
"title": "Multi-Modal Causal Disentanglement for Multi-Domain Sequential Audio Recommendation",
"line": 400
},
"DM2002": {
"group": "long-posters",
"floor": "3F",
"area": "Area B",
"slot": "P1-3B-048",
"session": "Heterogeneous data mining and systems",
"title": "HiMamba: A Hierarchical State Space Framework for Asymmetric Multimodal Data Mining in Clinical Assessment",
"line": 472
},
"DM2055": {
"group": "short-posters",
"floor": "3F",
"area": "Area A",
"slot": "P2-3A-038",
"session": "Applied Track: Real-world AI in finance, healthcare and industry I",
"title": "CGSBR: Adaptive Cross-Source Gating for Cold-Start Sequential Bundle Recommendation",
"line": 636
},
"DM2140": {
"group": "long-posters",
"floor": "3F",
"area": "Area A",
"slot": "P1-3A-008",
"session": "Machine learning and statistical methods",
"title": "Reliability-Guided Pseudo-Label Learning and Prototype Alignment for Domain-Adaptive Colorectal Segmentation",
"line": 432
},
"DM252": {
"group": "short-posters",
"floor": "2F",
"area": "Area B",
"slot": "P2-2B-046",
"session": "LLM-driven data mining and novel applications I",
"title": "Trust-But-Verify: Poisoning-Resilient Locally Private Graph Learning Protocols",
"line": 572
},
"DM383": {
"group": "long-posters",
"floor": "2F",
"area": "Area A",
"slot": "P1-2A-022",
"session": "Heterogeneous graphs and multimodal intelligence",
"title": "Subgraph Meta-Learning for Direction and Time Aware Transaction Anomaly Detection",
"line": 369
},
"DM463": {
"group": "short-posters",
"floor": "2F",
"area": "Area B",
"slot": "P2-2B-050",
"session": "Modeling, recommendation and LLM methods",
"title": "CaST-POI: Candidate-Conditioned Spatiotemporal Ranking for Next POI Recommendation",
"line": 576
},
"DM483": {
"group": "short-posters",
"floor": "2F",
"area": "Area B",
"slot": "P2-2B-051",
"session": "Modeling, recommendation and LLM methods",
"title": "Causal Masking Lowers Intrinsic Dimension in Full-Vocabulary Sequential Recommendation: Evidence from a Single-Variable Attention Ablation",
"line": 577
},
"DM713": {
"group": "short-posters",
"floor": "2F",
"area": "Area B",
"slot": "P2-2B-060",
"session": "Modeling, recommendation and LLM methods",
"title": "Hint-Conditioned Knowledge Tracing: A Causal-Inspired View of Context-Dependent Hint Usage",
"line": 586
},
"DM736": {
"group": "long-posters",
"floor": "3F",
"area": "Area B",
"slot": "P1-3B-050",
"session": "LLM-driven data mining and retrieval I",
"title": "From Natural-Language Hypotheses to Verified Data-Mining Discoveries",
"line": 474
},
"DM803": {
"group": "short-posters",
"floor": "2F",
"area": "Area B",
"slot": "P2-2B-032",
"session": "Heterogeneous data mining and multimodal learning I",
"title": "Graph-Guided LLM Reasoning for Incident Prediction Over Multi-Source Data Streams",
"line": 558
}
},
"oral_only": [],
"title_differences": [
{
"paper_id": "DM1319",
"poster_title": "FFE-FRCN: A Frequency Feature Enhanced Few-shot Learning Approach for Object Detection in Microscopic Images",
"oral_title": "An FFE-FRCN-Based Data Platform for Environmental Microorganism Image Retrieval and Knowledge Platform"
}
],
"nonpaper_table_rows": [],
"database_written": false,
"scope": "Official presentation metadata. Shared programme IDs identify repeated appearances within this event. Poster-only IDs are scheduling observations pending identity and publication review.",
"pg_readonly": {
"snapshot": {
"at": "2026-11-18 07:08:21.138511+00",
"snapshot": "1095530:1095530:"
},
"report_sha256": "04ad5547342d7e3c2939046f290ecff647878c202c867e47acbdc438d4e10ad4",
"existing_icdm2026_editions": [],
"existing_icdm_program_source": false,
"reviewed_links": [
{
"external_id": "2026/DM1125",
"paper_id": 2783475,
"track": "research",
"source_row_sha256": "4e981605cf471a5ea29bb3ba3a676784cea55f7777a681eb5fe267ff9c83fa4a",
"source_abstract": null,
"work_anchor": {
"scheme": "arxiv",
"value": "2608.29635"
},
"independent_source": "https://arxiv.org/abs/2608.29635",
"existing_history": [
{
"id": 132984,
"venue": "ICLR",
"year": 2026,
"observations": [
{
"id": 134279,
"status": "rejected"
},
{
"id": 539402,
"status": "rejected"
}
]
}
],
"original_icdm_decision_verified": false,
"imported": false
},
{
"external_id": "2026/DM790",
"paper_id": 2485633,
"track": "applied",
"source_row_sha256": "14e03ca3833b267d6f677325f4e4b9295d1fa42cc77be6707b9bca745444888b",
"source_abstract": null,
"work_anchor": {
"scheme": "arxiv",
"value": "2509.19774"
},
"independent_source": "https://arxiv.org/abs/2509.19774",
"existing_history": [
{
"id": 138980,
"venue": "ICLR",
"year": 2026,
"observations": [
{
"id": 545219,
"status": "withdrawn"
}
]
}
],
"original_icdm_decision_verified": false,
"imported": false
},
{
"external_id": "2026/DM1157",
"paper_id": 3286364,
"track": "research",
"source_row_sha256": "22ed027f29a161bc978b3c33e921b42c67d999d520eed49ec46ce5935a6f2f2a",
"source_abstract": null,
"work_anchor": {
"scheme": "openreview_snapshot",
"value": "2026/1dGzRTSLqL"
},
"independent_source": "https://dm.cs.univie.ac.at/news/story/deepsync-paper-accepted-at-icdm-2026",
"existing_history": [
{
"id": 127709,
"venue": "ICLR",
"year": 2026,
"observations": [
{
"id": 129094,
"status": "rejected"
},
{
"id": 534293,
"status": "rejected"
}
]
}
],
"original_icdm_decision_verified": false,
"imported": false
}
],
"held": [
{
"external_id": "2026/DM1791",
"paper_id": 3288747,
"reason": "DomainBridging has name-order/format differences and no arxiv self-accepted link; token-bag retrieval is not author or work identity approval."
},
{
"external_id": "2026/DM1801",
"paper_id": 3288866,
"reason": "LightGFC author lists substantially differ; hold version/work review, no automatic author aliases or merge."
},
{
"external_id": "2026/DM1413",
"paper_id": 3292171,
"reason": "OFS author lists differ including JianchengLv vs JindiLv; not treated as aliases; hold independent version/work verification."
}
]
},
"publisher_catalog_status": "No 2026 IEEE publisher catalogue verified in this check",
"public_negative_availability": "unverified",
"next": "Use the existing oral source denominator by track. Compare the three reviewed candidates with the fresh import baseline and preserve their earlier attempts. Keep the poster-only rows as a separate source inventory, and resolve the title disagreement before linking that poster."
}
-10
View File
@@ -44,13 +44,3 @@ paper-atlas collect icdm2026-program \
原始来源恢复包为 `/data/paper-index/validation/remaining2026-official-leads-v1-recovery.tar.gz`,SHA-256 为 `0be8c84bb5c591e5afaaa5f95dd5334bf13a9be1671cb942e46e998f502495bb`。独立盘点为 `/data/paper-index/validation/icdm2026-independent-program-inventory-v1.json`,SHA-256 为 `e13eaea0a158dcaafdf5dfb4af7948e4793b0e98041ec87d42b52513acfdc5f6`;本机工作记录目录也保留同名备份。 原始来源恢复包为 `/data/paper-index/validation/remaining2026-official-leads-v1-recovery.tar.gz`,SHA-256 为 `0be8c84bb5c591e5afaaa5f95dd5334bf13a9be1671cb942e46e998f502495bb`。独立盘点为 `/data/paper-index/validation/icdm2026-independent-program-inventory-v1.json`,SHA-256 为 `e13eaea0a158dcaafdf5dfb4af7948e4793b0e98041ec87d42b52513acfdc5f6`;本机工作记录目录也保留同名备份。
下次继续先验证源码版本、原始证据和实际输出,再寻找可公开取得的摘要与正式出版记录,核查逐篇版本、稳定标识及旧投稿历史,准备新的只读预检和可审阅入库队列。完成日程规范化不意味着 ICDM2026 整届出版覆盖、公开负面决定普查或最终全库去重完成。 下次继续先验证源码版本、原始证据和实际输出,再寻找可公开取得的摘要与正式出版记录,核查逐篇版本、稳定标识及旧投稿历史,准备新的只读预检和可审阅入库队列。完成日程规范化不意味着 ICDM2026 整届出版覆盖、公开负面决定普查或最终全库去重完成。
## 2026-11-18 官方展示来源核查
口头日程的 324 个节目编号全部出现在官方海报排期。海报表有 343 个不同编号,其中长演讲组安排 177 项海报,短演讲组安排 166 项。19 个只见于海报表的编号作为独立来源线索登记。两份排期的覆盖范围各自保留,正式出版目录仍待核验。[官方详细日程](https://icdm2026.neu.edu.cn/DetailedSessions/list.htm) · [官方海报表](https://www3.cs.stonybrook.edu/~icdm2026/PosterwDemo/list.htm)
共同编号 DM1319 的标题在两个来源中实质不同。详细日程描述基于 FFE-FRCN 的数据平台,海报表描述 FFE-FRCN 的少样本检测方法。两份标题及来源位置均已留存,该编号的跨来源作品关联待核验。
生产只读快照确认 ICDM 2026 仍无入库届次。节目、作者和独立录用证据支持继续处理 DM1125、DM790、DM1157 三项既有作品关联。DM1125 和 DM1157 各有一次 ICLR 拒稿投稿及两条来源观察,DM790 有一次 ICLR 撤回投稿。三项均尚未导入,节目原始摘要字段缺失。另有三个候选存在作者名称顺序或名单差异,继续暂缓作品关联。[FALCON 作者记录](https://arxiv.org/abs/2608.29635) · [PPGFlowECG 作者记录](https://arxiv.org/abs/2509.19774) · [DeepSynC 大学录用说明](https://dm.cs.univie.ac.at/news/story/deepsync-paper-accepted-at-icdm-2026)
本轮仍未核实公开的 ICDM 2026 拒稿或撤回决定导出。[逐条差集、标题冲突、快照和后续入口](coverage/icdm2026-presentation-scope-2026-11-18.json)
+1
View File
@@ -28,6 +28,7 @@ COMMANDS = {
('collect','eccv2026-program'): ('paper_atlas.collectors.eccv2026', {}, '核验ECCV2026暂定节目并断点采集公开元数据'), ('collect','eccv2026-program'): ('paper_atlas.collectors.eccv2026', {}, '核验ECCV2026暂定节目并断点采集公开元数据'),
('collect','icassp2026-registry'): ('paper_atlas.collectors.icassp2026', {}, '核验 ICASSP 2026 冻结注册论文元数据'), ('collect','icassp2026-registry'): ('paper_atlas.collectors.icassp2026', {}, '核验 ICASSP 2026 冻结注册论文元数据'),
('collect','icdm2026-program'): ('paper_atlas.collectors.icdm2026', {}, '核验 ICDM2026 冻结日程及明确的研究/应用轨道'), ('collect','icdm2026-program'): ('paper_atlas.collectors.icdm2026', {}, '核验 ICDM2026 冻结日程及明确的研究/应用轨道'),
('collect','cikm2026-sources'): ('paper_atlas.collectors.cikm2026', {}, '核对 CIKM 2026 节目、附属议程与 Crossref 出版登记'),
('collect','sdm2026-proceedings'): ('paper_atlas.collectors.sdm2026', {}, '核验 SDM 2026 注册论文并单独保留公开日程版本'), ('collect','sdm2026-proceedings'): ('paper_atlas.collectors.sdm2026', {}, '核验 SDM 2026 注册论文并单独保留公开日程版本'),
('collect','icaps2026-program'): ('paper_atlas.collectors.icaps2026', {}, '核验 ICAPS 2026 官方录用名单并分开既有论文展示'), ('collect','icaps2026-program'): ('paper_atlas.collectors.icaps2026', {}, '核验 ICAPS 2026 官方录用名单并分开既有论文展示'),
('collect','rss2026-proceedings'): ('paper_atlas.collectors.rss2026', {}, '核验 RSS 2026 官方论文集与独立录用表'), ('collect','rss2026-proceedings'): ('paper_atlas.collectors.rss2026', {}, '核验 RSS 2026 官方论文集与独立录用表'),
+164
View File
@@ -0,0 +1,164 @@
"""Collect CIKM 2026 programme and Crossref metadata from frozen snapshots."""
import argparse
from collections import Counter
from hashlib import sha256
import json
from pathlib import Path
import re
from paper_atlas.io.html import Document, clean
PROGRAM = 'https://cikm2026.diag.uniroma1.it/cikm-program/program.html'
VOLUME = '10.1145/3799682'
CONTAINER = 'Proceedings of the 35th ACM International Conference on Information and Knowledge Management'
TRACKS = {'FULL': 596, 'SHORT': 234, 'APPLIED': 86, 'RESOURCE': 46, 'DEMO': 46}
SUPPLEMENTAL = {'KEYNOTE', 'TUTORIAL', 'INDUSTRY', 'POSTER', 'PHD', 'PHD · REMOTE', 'WORKSHOP', 'ANALYTICUP'}
PAGE_COUNTS = (250, 250, 250, 250, 81)
HASHES = {
'program.html': '3eaa668b6aaf09a671d8a3e19a2889b9ab21afaa7ded013794afa94bf35281d6',
'CIKM2026_program.html': 'e3630bd1247df713b8b618f1d3cd9709e7020b2524bb677f82a10f83badd1732',
'crossref-parent.json': 'f3370418d86ccfcfaeb972073bc5fcd1c1c737caed96570fc83f99db0f7d204a',
'crossref-page-1.json': '16bed29f8ee82d668b124c369b632c5c1de0753bdbec7eb72ecb22ffdda5c710',
'crossref-page-2.json': '54424ca31d99462394c9fde5fecfd7dd16337351d21ede3169f19af54d4f49c5',
'crossref-page-3.json': '2dc31d89f41966614ed66a11d1ebaa4b2a6904dcf92945b7185e72e7e65382dd',
'crossref-page-4.json': 'fbbec2b52650dc99a0251c743f035a6fd95be263b04c55c12134cf8473bdde76',
'crossref-page-5.json': '06151c5f770c1e37e8ea9dabe5210be9c969a14fa3bf9925a2d7730ced47d6e5',
'crossref-pages.json': '8be1f8f05139f7a1231b73ad6430f28d2f625809f1530cd9de654f8b3c68c1ae',
}
def fields(node, css_class):
return [child for child in node.walk() if css_class in child.attrs.get('class', '').split()]
def parse_program(raw, *, require_doi):
document = Document(raw.decode('utf-8'))
papers = {}
for card in fields(document.root, 'paper'):
pid = card.attrs.get('data-pid')
if not pid:
continue
titles = fields(card, 'ptitle')
labels = fields(card, 'badge')
abstracts = fields(card, 'abstxt')
if pid in papers or len(titles) != 1 or len(labels) != 1 or len(abstracts) > 1:
raise ValueError(f'Invalid or repeated programme card {pid}')
title = clean(titles[0].text())
label = clean(labels[0].text())
authors = [clean(author.text()) for author in fields(card, 'au')]
abstract = clean(abstracts[0].text()) if abstracts else None
doi_url = titles[0].attrs.get('href')
doi = doi_url.removeprefix('https://doi.org/') if doi_url else None
if label not in TRACKS or not title or not authors or not all(authors):
raise ValueError(f'Incomplete programme metadata {pid}')
if require_doi and (not doi or not re.fullmatch(re.escape(VOLUME) + r'\.\d+', doi) or not abstract):
raise ValueError(f'Missing programme DOI or abstract {pid}')
papers[pid] = {'programme_id': pid, 'title': title, 'author_display_names': authors,
'abstract': abstract, 'programme_label': label, 'programme_doi_link': doi_url,
'doi': doi if require_doi else None}
return papers
def parse_supplemental(raw):
document = Document(raw.decode('utf-8'))
events = {}
for session in fields(document.root, 'sessblock'):
section = session.attrs.get('id', '')
if not section.startswith(('pre-', 'kn-')):
continue
for card in fields(session, 'paper'):
labels = fields(card, 'badge')
if len(labels) != 1:
raise ValueError(f'Ambiguous supplemental label in {section}')
label = clean(labels[0].text())
if label not in SUPPLEMENTAL:
continue
titles = fields(card, 'ptitle')
abstracts = fields(card, 'abstxt')
locator = card.attrs.get('id') or section
if locator in events or len(titles) != 1 or len(abstracts) > 1:
raise ValueError(f'Ambiguous supplemental entry {locator}')
events[locator] = {'programme_locator': locator, 'programme_section': section,
'programme_label': label, 'title': clean(titles[0].text()),
'author_display_names': [clean(a.text()) for a in fields(card, 'au')],
'abstract': clean(abstracts[0].text()) if abstracts else None}
return events
def parse_registry(snapshots):
volume = json.loads(snapshots['crossref-parent.json'])['message']
if (volume['DOI'], volume['title'], volume['ISBN']) != (VOLUME, [CONTAINER], ['9798400725395']):
raise ValueError('Unexpected Crossref volume')
records = {}
for number, count in enumerate(PAGE_COUNTS, 1):
name = f'crossref-page-{number}.json'
message = json.loads(snapshots[name])['message']
if message['total-results'] != 1081 or len(message['items']) != count:
raise ValueError(f'Unexpected Crossref pagination in {name}')
for item in message['items']:
doi = item['DOI'].lower()
if (doi in records or not re.fullmatch(re.escape(VOLUME) + r'\.\d+', doi)
or item.get('container-title') != [CONTAINER] or item.get('type') != 'proceedings-article'
or item.get('published', {}).get('date-parts') != [[2026, 11, 6]]):
raise ValueError(f'Unexpected Crossref record {doi}')
records[doi] = {'registry': item, 'source_file': name}
return records
def collect(snapshots, output):
snapshots = Path(snapshots)
output = Path(output)
if output.exists() or output.is_symlink():
raise FileExistsError(output)
bodies = {}
for name, expected in HASHES.items():
path = snapshots / name
if path.is_symlink() or not path.is_file() or path.stat().st_size > 8 * 1024**2:
raise ValueError(f'Invalid snapshot file {name}')
raw = path.read_bytes()
if sha256(raw).hexdigest() != expected:
raise ValueError(f'Snapshot hash mismatch for {name}')
bodies[name] = raw
programme = parse_program(bodies['program.html'], require_doi=True)
alternate = parse_program(bodies['CIKM2026_program.html'], require_doi=False)
supplemental = parse_supplemental(bodies['program.html'])
registry = parse_registry(bodies)
programme_dois = {r['doi'] for r in programme.values()}
tracks = Counter(r['programme_label'] for r in programme.values())
if (tracks != TRACKS or len(programme_dois) != len(programme) or programme_dois - registry.keys()
or len(alternate) != 1005 or alternate.keys() - programme.keys()
or len(supplemental) != 73 or len(registry) - len(programme_dois) != 73):
raise ValueError('Unexpected programme or registry coverage')
groups = {
'programme': programme,
'alternate-programme': alternate,
'supplemental-programme': supplemental,
'crossref': registry,
}
output.mkdir(parents=True, exist_ok=False)
files = {}
for name, rows in groups.items():
filename = name + '.jsonl'
payload = ''.join(json.dumps({'source_key': key, **row}, ensure_ascii=False) + '\n'
for key, row in sorted(rows.items())).encode()
(output / filename).write_bytes(payload)
files[filename] = {'records': len(rows), 'sha256': sha256(payload).hexdigest()}
report = {'status': 'collected_pending_identity_review', 'files': files, 'input_sha256': HASHES,
'programme_by_label': dict(tracks), 'programme_dois_in_registry': len(programme_dois),
'registry_outside_doi_programme': sorted(registry.keys() - programme_dois),
'programme_ids_missing_from_alternate': sorted(programme.keys() - alternate.keys()),
'database_written': False, 'public_negative_decisions_verified': False}
partial = output / 'collection-report.json.part'
partial.write_text(json.dumps(report, ensure_ascii=False, indent=2) + '\n', encoding='utf-8')
partial.replace(output / 'collection-report.json')
return report
def main(argv=None, *, prog=None):
parser = argparse.ArgumentParser(prog=prog, description=__doc__)
parser.add_argument('--snapshots', type=Path, required=True)
parser.add_argument('--out', type=Path, required=True)
options = parser.parse_args(argv)
print(json.dumps(collect(options.snapshots, options.out), ensure_ascii=False))