-
Notifications
You must be signed in to change notification settings - Fork 1
Expand file tree
/
Copy pathawesome-ai-auto-research.html
More file actions
5030 lines (4679 loc) · 361 KB
/
Copy pathawesome-ai-auto-research.html
File metadata and controls
5030 lines (4679 loc) · 361 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
<!doctype html>
<html lang="en">
<head>
<title>AI for Auto-Research: A Survey</title>
<link rel="icon" href="assets/auto-research/figures/avatar.png" type="image/png">
<link rel="preconnect" href="https://api.fontshare.com" crossorigin>
<link rel="preconnect" href="https://cdn.fontshare.com" crossorigin>
<link rel="preconnect" href="https://fonts.googleapis.com">
<link rel="preconnect" href="https://fonts.gstatic.com" crossorigin>
<link href="https://api.fontshare.com/v2/css?f[]=satoshi@300,400,500,600,700,800&display=swap" rel="stylesheet">
<link href="https://fonts.googleapis.com/css2?family=Fraunces:opsz,wght@9..144,300;9..144,400;9..144,500;9..144,600;9..144,700&display=swap" rel="stylesheet">
<meta charset="utf-8">
<meta name="viewport" content="width=device-width, initial-scale=1">
<meta name="description" content="AI for Auto-Research — the first survey of AI across the complete research lifecycle, covering idea generation, literature review, coding, writing, peer review, rebuttal, and dissemination.">
<script src="./static/js/distill_template.v2.js"></script>
<link rel="stylesheet" href="./static/css/style.css">
<link rel="stylesheet" href="./static/css/auto-research.css">
<link rel="stylesheet" href="./static/css/fontawesome.all.min.css">
<link rel="stylesheet" href="https://cdn.jsdelivr.net/gh/jpswalsh/academicons@1/css/academicons.min.css">
<script defer src="./static/js/fontawesome.all.min.js"></script>
<script src="https://cdn.jsdelivr.net/npm/jquery@3.7.1/dist/jquery.min.js"></script>
<script defer src="./static/js/medium-zoom.min.js"></script>
<script defer src="./static/js/zoom.js"></script>
</head>
<body>
<div class="scroll-progress" id="scrollProgress" style="background: linear-gradient(90deg, #92400e, #d97706, #f59e0b, #fbbf24, #fde68a);"></div>
<div class="header-wrapper" style="background: linear-gradient(220deg, #fde68a 0%, #92400ec0 40%, #1c1917 100%);">
<div class="header-container" id="header-container">
<div class="header-content">
<h1 style="font-size:4.0rem;">
AI for Auto-Research: A Survey
</h1>
<h2 style="margin:1.2rem 0 1.6rem; font-size:1.05rem; line-height:1.55; color:#FFF7D4; font-weight:400; max-width:540px;">
<span style="display:flex; align-items:center; gap:0.4rem; margin-bottom:1.2rem; font-weight:600; font-size:1.5rem; color:#fff;">
Awesome AI Auto-Research Team
</span>
<span style="opacity:0.92;">A comprehensive practitioner's guide for using AI-assisted tools across the complete academic research lifecycle.</span>
</h2>
<div class="button-container">
<a href="https://arxiv.org/abs/2605.18661" class="button paper-link" target="_blank">
<span class="icon is-small">
<i class="ai ai-arxiv" style="height: 1.5em;"></i>
</span><span>arXiv</span>
</a>
<a href="https://worldbench.github.io/assets_common/papers/survey-ai-auto-research.pdf" class="button paper-link" target="_blank">
<span class="icon is-small">
<img src="assets_common/icons/pdf.png" style="height: 1.4em;">
</span><span>PDF</span>
</a>
<a href="https://github.com/worldbench/awesome-auto-research" class="button" target="_blank">
<span class="icon is-small">
<img src="assets_common/icons/github.png" style="height: 1.4em;">
</span><span>GitHub</span>
</a>
<a href="https://huggingface.co/papers/2605.18661" class="button" target="_blank">
<span class="icon is-small">
<img src="assets_common/icons/hf.png" style="height: 1.5em;">
</span><span>HuggingFace Paper</span>
</a>
</div>
</div>
<div class="header-image">
<img src="assets/auto-research/figures/teaser.png" alt="AI Auto-Research Teaser" class="teaser-image">
</div>
</div>
</div>
<d-article>
<d-contents>
<nav>
<h4>Contents</h4>
<div><a href="#abstract">Abstract</a></div>
<div><a href="#lifecycle">Research Lifecycle</a></div>
<div><a href="#collection">Paper Collection</a></div>
<div><a href="#exhibition">Exhibition</a></div>
<div><a href="#citation">Citation</a></div>
<div><a href="#responsible">Responsible Use</a></div>
<div><a href="#authors">Authors</a></div>
</nav>
</d-contents>
<!-- ==================== Abstract ==================== -->
<section id="abstract">
<div class="wb-section">
<!-- Hero pull-quote: capability vs integrity tension -->
<div class="ar-hero">
<div class="ar-hero-tag">Crossing a threshold</div>
<p class="ar-hero-quote">
AI-assisted research is crossing a threshold. Fully automated systems can now generate research papers for as little as <b>$15</b>, while long-horizon agents execute experiments, draft manuscripts, and simulate critique with minimal human input — yet under scientific pressure, even frontier LLMs still fabricate experiment results, miss hidden errors, and fail to judge novelty reliably.
</p>
</div>
<!-- Teaser figure -->
<figure class="ar-teaser-figure">
<img src="assets/auto-research/figures/teaser_paper.png" alt="AI auto-research across the complete lifecycle: four phases and eight stages.">
<figcaption class="ar-teaser-caption">
<b>AI auto-research across the complete lifecycle.</b> Four phases and eight stages: <b>Creation</b> (ideation, literature, code & experiments, tables & figures), <b>Writing</b> (paper writing), <b>Validation</b> (peer review, rebuttal & revision), and <b>Dissemination</b> (posters, slides, videos, social media, project pages, and interactive paper agents).
</figcaption>
</figure>
<p class="ar-lead">
Across this lifecycle we identify a <b>sharp, stage-dependent boundary</b> between reliable assistance and unreliable autonomy. The core challenge is no longer whether AI can produce the <i>forms</i> of research, but whether it can preserve the <i>substance</i> — evidence, judgment, provenance, and accountability. We argue that <b>human-governed collaboration</b> is the most credible deployment paradigm, and that effective systems converge on layered architectures combining <b>exploration, execution, and verification</b>.
</p>
<div class="ar-stats">
<div class="ar-stat">
<div class="ar-stat-num">250+</div>
<div class="ar-stat-label">Papers Surveyed</div>
</div>
<div class="ar-stat">
<div class="ar-stat-num">8</div>
<div class="ar-stat-label">Research Stages</div>
</div>
<div class="ar-stat">
<div class="ar-stat-num">4</div>
<div class="ar-stat-label">Lifecycle Phases</div>
</div>
<div class="ar-stat">
<div class="ar-stat-num">5</div>
<div class="ar-stat-label">Central Findings</div>
</div>
</div>
<br><br>
</div>
</section>
<!-- ==================== Research Lifecycle ==================== -->
<section id="lifecycle">
<div class="wb-section">
<h2>The Academic Research Lifecycle</h2>
<div class="wb-kicker">
Four Phases, Eight Stages
</div>
<div class="ar-hr"></div>
<p class="ar-lead">
We organize the academic research lifecycle as <b>eight interconnected stages</b> grouped into <b>four epistemological phases</b>. Each phase serves a distinct function in producing, scrutinizing, and communicating scientific knowledge.
</p>
<div class="lifecycle-blocks">
<!-- ====== Phase 1: Creation ====== -->
<div class="phase-block pc1">
<div class="phase-header-card">
<div class="phase-hero-image">
<img src="assets/auto-research/figures/phase1.png" alt="Phase 1: Creation">
<div class="phase-hero-overlay">
<div class="phase-hero-roman">Phase 1 · Stages 1–4</div>
<h3 class="phase-hero-title">Creation</h3>
<p class="phase-hero-subtitle">The stages through which a research contribution is materially produced.</p>
</div>
</div>
<div class="phase-motivation">
<p>From <b>spark to substance</b>: ideation grounded in the literature, executed in code, and rendered as figures and tables. This is where AI's promise — and its capability boundary — is most visible.</p>
</div>
</div>
<div class="stage-card-grid">
<article class="stage-card" data-stage="s1" onclick="toggleStage(this)">
<div class="stage-image"><img src="assets/auto-research/figures/stages/s1.png" alt="S1: Idea Generation"></div>
<div class="stage-body">
<div class="stage-tag">Stage 1</div>
<h4 class="stage-name">Idea Generation</h4>
<span class="stage-more">Know more</span>
</div>
<div class="stage-detail-content">
<p class="stage-detail-desc">Generating, refining, and evaluating <b>research hypotheses</b>. Systems span direct LLM prompting, retrieval-augmented and knowledge-graph generation, multi-agent collaboration, and learned quality signals. The central challenge: LLMs can produce ideas that appear novel and well-motivated, yet often struggle to generate ones that remain <b>feasible, distinctive, and impactful</b> after execution.</p>
<div class="stage-detail-section">
<div class="stage-detail-section-label">Sub-topics</div>
<div class="stage-detail-pills">
<span class="stage-detail-pill">LLM-based generation</span>
<span class="stage-detail-pill">KG-driven</span>
<span class="stage-detail-pill">Trend-driven</span>
<span class="stage-detail-pill">Multi-agent ideation</span>
<span class="stage-detail-pill">Novelty assessment</span>
<span class="stage-detail-pill">Human-AI co-ideation</span>
</div>
</div>
<div class="stage-detail-section">
<div class="stage-detail-section-label">Representative methods</div>
<div class="stage-detail-methods">
<span class="stage-detail-method">AI Scientist</span>
<span class="stage-detail-method">ResearchAgent</span>
<span class="stage-detail-method">SciMON</span>
<span class="stage-detail-method">VirSci</span>
<span class="stage-detail-method">SciAgents</span>
<span class="stage-detail-method">IdeaSynth</span>
<span class="stage-detail-method">MOOSE-Chem</span>
</div>
</div>
</div>
</article>
<article class="stage-card" data-stage="s2" onclick="toggleStage(this)">
<div class="stage-image"><img src="assets/auto-research/figures/stages/s2.png" alt="S2: Literature Review"></div>
<div class="stage-body">
<div class="stage-tag">Stage 2</div>
<h4 class="stage-name">Literature Review</h4>
<span class="stage-more">Know more</span>
</div>
<div class="stage-detail-content">
<p class="stage-detail-desc">Retrieving, synthesizing, and organizing prior work into <b>coherent research contexts</b>. Compared with idea generation, this stage is more grounded and externally verifiable, making it one of the <b>fastest-maturing</b> areas in AI-assisted research. Yet faithful citation, coverage completeness, and multi-paper relational reasoning remain difficult.</p>
<div class="stage-detail-section">
<div class="stage-detail-section-label">Sub-topics</div>
<div class="stage-detail-pills">
<span class="stage-detail-pill">Semantic retrieval</span>
<span class="stage-detail-pill">Survey generation</span>
<span class="stage-detail-pill">Deep Research</span>
<span class="stage-detail-pill">Citation graph</span>
<span class="stage-detail-pill">Hierarchical org.</span>
<span class="stage-detail-pill">Related work gen.</span>
</div>
</div>
<div class="stage-detail-section">
<div class="stage-detail-section-label">Representative methods</div>
<div class="stage-detail-methods">
<span class="stage-detail-method">PaperQA2</span>
<span class="stage-detail-method">AutoSurvey</span>
<span class="stage-detail-method">STORM</span>
<span class="stage-detail-method">SurveyX</span>
<span class="stage-detail-method">SurveyForge</span>
<span class="stage-detail-method">OpenScholar</span>
</div>
</div>
</div>
</article>
<article class="stage-card" data-stage="s3" onclick="toggleStage(this)">
<div class="stage-image"><img src="assets/auto-research/figures/stages/s3.png" alt="S3: Coding & Experiments"></div>
<div class="stage-body">
<div class="stage-tag">Stage 3</div>
<h4 class="stage-name">Coding & Experiments</h4>
<span class="stage-more">Know more</span>
</div>
<div class="stage-detail-content">
<p class="stage-detail-desc">Translating ideas into <b>executable code</b>, running experiments, and analyzing empirical results. The challenge is not whether LLMs can write plausible code, but whether they can produce <b>semantically correct</b> research implementations, execute meaningful experiments, and interpret results reliably — performance still drops sharply on genuinely novel research code.</p>
<div class="stage-detail-section">
<div class="stage-detail-section-label">Sub-topics</div>
<div class="stage-detail-pills">
<span class="stage-detail-pill">Paper-to-Code</span>
<span class="stage-detail-pill">Experiment design</span>
<span class="stage-detail-pill">Auto execution</span>
<span class="stage-detail-pill">Result analysis</span>
<span class="stage-detail-pill">Lab automation</span>
<span class="stage-detail-pill">Benchmarking</span>
</div>
</div>
<div class="stage-detail-section">
<div class="stage-detail-section-label">Representative methods</div>
<div class="stage-detail-methods">
<span class="stage-detail-method">PaperCoder</span>
<span class="stage-detail-method">MLAgentBench</span>
<span class="stage-detail-method">AIDE</span>
<span class="stage-detail-method">R&D-Agent</span>
<span class="stage-detail-method">ChemCrow</span>
<span class="stage-detail-method">Coscientist</span>
</div>
</div>
</div>
</article>
<article class="stage-card" data-stage="s4" onclick="toggleStage(this)">
<div class="stage-image"><img src="assets/auto-research/figures/stages/s4.png" alt="S4: Tables & Figures"></div>
<div class="stage-body">
<div class="stage-tag">Stage 4</div>
<h4 class="stage-name">Tables & Figures</h4>
<span class="stage-more">Know more</span>
</div>
<div class="stage-detail-content">
<p class="stage-detail-desc">Constructing <b>method diagrams, result plots, comparison tables</b>, mathematical formulas, and algorithmic illustrations. Despite their importance in daily research practice, this stage remains <b>comparatively underdeveloped</b> — current tools serve as assistants rather than autonomous producers, and AI-generated figures frequently require human modification for domain-specific symbols and paper-specific visual languages.</p>
<div class="stage-detail-section">
<div class="stage-detail-section-label">Sub-topics</div>
<div class="stage-detail-pills">
<span class="stage-detail-pill">Method diagrams</span>
<span class="stage-detail-pill">Result plots</span>
<span class="stage-detail-pill">Table generation</span>
<span class="stage-detail-pill">LaTeX / TikZ</span>
<span class="stage-detail-pill">SVG generation</span>
<span class="stage-detail-pill">Visual feedback</span>
</div>
</div>
<div class="stage-detail-section">
<div class="stage-detail-section-label">Representative methods</div>
<div class="stage-detail-methods">
<span class="stage-detail-method">AutoFigure</span>
<span class="stage-detail-method">MatPlotAgent</span>
<span class="stage-detail-method">PlotGen</span>
<span class="stage-detail-method">ChartGPT</span>
<span class="stage-detail-method">AutomaTikZ</span>
<span class="stage-detail-method">DeTikZify</span>
</div>
</div>
</div>
</article>
</div>
<div class="stage-detail-panel"><div class="stage-detail-inner"></div></div>
</div>
<!-- ── Phase 1 → 2 separator ── -->
<div class="phase-separator"></div>
<!-- ====== Phase 2: Writing ====== -->
<div class="phase-block pc2">
<div class="phase-header-card">
<div class="phase-hero-image">
<img src="assets/auto-research/figures/phase2.png" alt="Phase 2: Writing">
<div class="phase-hero-overlay">
<div class="phase-hero-roman">Phase 2 · Stage 5</div>
<h3 class="phase-hero-title">Writing</h3>
<p class="phase-hero-subtitle">Organizing the outputs of Creation into a formal scholarly manuscript for communication and external scrutiny.</p>
</div>
</div>
<div class="phase-motivation">
<p>Writing is not a formatting step — it is a <b>rhetorical and evidential organization</b> process that requires distinct AI capabilities from those used to produce code, experiments, or figures.</p>
</div>
</div>
<div class="stage-card-grid">
<article class="stage-card" data-stage="s5" onclick="toggleStage(this)">
<div class="stage-image"><img src="assets/auto-research/figures/stages/s5.png" alt="S5: Paper Writing"></div>
<div class="stage-body">
<div class="stage-tag">Stage 5</div>
<h4 class="stage-name">Paper Writing</h4>
<span class="stage-more">Know more</span>
</div>
<div class="stage-detail-content">
<p class="stage-detail-desc">Drafting, editing, polishing, and structuring academic manuscripts. AI assistance ranges from <b>grammar correction</b> and citation support to section-level drafting and full-paper generation. The central failure mode is no longer ungrammatical prose but the gap between fluency and <b>argumentative depth</b>; end-to-end autonomous systems have not yet consistently reached major-venue acceptance standards.</p>
<div class="stage-detail-section">
<div class="stage-detail-section-label">Sub-topics</div>
<div class="stage-detail-pills">
<span class="stage-detail-pill">Semi-auto assist</span>
<span class="stage-detail-pill">Full-auto generation</span>
<span class="stage-detail-pill">Citation insertion</span>
<span class="stage-detail-pill">AI text detection</span>
<span class="stage-detail-pill">LaTeX generation</span>
<span class="stage-detail-pill">Style & polishing</span>
</div>
</div>
<div class="stage-detail-section">
<div class="stage-detail-section-label">Representative methods</div>
<div class="stage-detail-methods">
<span class="stage-detail-method">CycleResearcher</span>
<span class="stage-detail-method">ScholarCopilot</span>
<span class="stage-detail-method">CoAuthor</span>
<span class="stage-detail-method">TeXpert</span>
<span class="stage-detail-method">FutureGen</span>
<span class="stage-detail-method">FARS</span>
</div>
</div>
</div>
</article>
</div>
<div class="stage-detail-panel"><div class="stage-detail-inner"></div></div>
</div>
<!-- ── Phase 2 → 3 separator ── -->
<div class="phase-separator"></div>
<!-- ====== Phase 3: Validation ====== -->
<div class="phase-block pc3">
<div class="phase-header-card">
<div class="phase-hero-image">
<img src="assets/auto-research/figures/phase3.png" alt="Phase 3: Validation">
<div class="phase-hero-overlay">
<div class="phase-hero-roman">Phase 3 · Stages 6–7</div>
<h3 class="phase-hero-title">Validation</h3>
<p class="phase-hero-subtitle">External scrutiny — how the community challenges and refines a manuscript.</p>
</div>
</div>
<div class="phase-motivation">
<p>Together, peer review and rebuttal form the <b>community-facing mechanism</b> through which scientific claims are challenged, defended, and revised — not a bureaucratic afterthought, but the verification layer of research.</p>
</div>
</div>
<div class="stage-card-grid">
<article class="stage-card" data-stage="s6" onclick="toggleStage(this)">
<div class="stage-image"><img src="assets/auto-research/figures/stages/s6.png" alt="S6: Peer Review"></div>
<div class="stage-body">
<div class="stage-tag">Stage 6</div>
<h4 class="stage-name">Peer Review</h4>
<span class="stage-more">Know more</span>
</div>
<div class="stage-detail-content">
<p class="stage-detail-desc">Generating <b>structured reviews</b>, matching reviewers to manuscripts, assessing review quality, and supporting meta-review decisions. These systems aim to <i>assist</i>, not replace, the community's evaluative process — standalone AI review remains <b>unsafe</b>, with persistent leniency bias and vulnerability to adversarial manipulation.</p>
<div class="stage-detail-section">
<div class="stage-detail-section-label">Sub-topics</div>
<div class="stage-detail-pills">
<span class="stage-detail-pill">Review generation</span>
<span class="stage-detail-pill">Meta-review</span>
<span class="stage-detail-pill">Reviewer matching</span>
<span class="stage-detail-pill">Quality assessment</span>
<span class="stage-detail-pill">Bias detection</span>
<span class="stage-detail-pill">Review feedback</span>
</div>
</div>
<div class="stage-detail-section">
<div class="stage-detail-section-label">Representative methods</div>
<div class="stage-detail-methods">
<span class="stage-detail-method">DeepReviewer</span>
<span class="stage-detail-method">MARG</span>
<span class="stage-detail-method">REMOR</span>
<span class="stage-detail-method">AgentReview</span>
<span class="stage-detail-method">OpenReviewer</span>
</div>
</div>
</div>
</article>
<article class="stage-card" data-stage="s7" onclick="toggleStage(this)">
<div class="stage-image"><img src="assets/auto-research/figures/stages/s7.png" alt="S7: Rebuttal & Revision"></div>
<div class="stage-body">
<div class="stage-tag">Stage 7</div>
<h4 class="stage-name">Rebuttal & Revision</h4>
<span class="stage-more">Know more</span>
</div>
<div class="stage-detail-content">
<p class="stage-detail-desc">Analyzing reviewer comments, identifying required evidence, drafting <b>evidence-grounded responses</b>, and supporting manuscript revision. This stage connects external critique with additional analysis, clarification, and experimental follow-up — <b>under-served relative to its practical importance</b>, despite being where authors directly negotiate reviewer concerns before final decisions.</p>
<div class="stage-detail-section">
<div class="stage-detail-section-label">Sub-topics</div>
<div class="stage-detail-pills">
<span class="stage-detail-pill">Comment analysis</span>
<span class="stage-detail-pill">Response generation</span>
<span class="stage-detail-pill">Evidence retrieval</span>
<span class="stage-detail-pill">Strategy planning</span>
<span class="stage-detail-pill">Effectiveness analysis</span>
</div>
</div>
<div class="stage-detail-section">
<div class="stage-detail-section-label">Representative methods</div>
<div class="stage-detail-methods">
<span class="stage-detail-method">RebuttalAgent</span>
<span class="stage-detail-method">Paper2Rebuttal</span>
<span class="stage-detail-method">ReviewerToo</span>
<span class="stage-detail-method">ReviewMT</span>
<span class="stage-detail-method">Re²</span>
<span class="stage-detail-method">Author-in-Loop</span>
</div>
</div>
</div>
</article>
</div>
<div class="stage-detail-panel"><div class="stage-detail-inner"></div></div>
</div>
<!-- ── Phase 3 → 4 separator ── -->
<div class="phase-separator"></div>
<!-- ====== Phase 4: Dissemination ====== -->
<div class="phase-block pc4">
<div class="phase-header-card">
<div class="phase-hero-image">
<img src="assets/auto-research/figures/phase4.png" alt="Phase 4: Dissemination">
<div class="phase-hero-overlay">
<div class="phase-hero-roman">Phase 4 · Stage 8</div>
<h3 class="phase-hero-title">Dissemination</h3>
<p class="phase-hero-subtitle">Translating the manuscript into formats accessible to broader audiences.</p>
</div>
</div>
<div class="phase-motivation">
<p>Posters, slides, videos, project pages, social-media summaries, <b>and interactive agents</b> are independent knowledge artifacts with their own fidelity and trust requirements — not appendages to the paper.</p>
</div>
</div>
<div class="stage-card-grid">
<article class="stage-card" data-stage="s8" onclick="toggleStage(this)">
<div class="stage-image"><img src="assets/auto-research/figures/stages/paper2x.png" alt="S8: Paper2X"></div>
<div class="stage-body">
<div class="stage-tag">Stage 8</div>
<h4 class="stage-name">Paper2X</h4>
<span class="stage-more">Know more</span>
</div>
<div class="stage-detail-content">
<p class="stage-detail-desc">Converting papers into <b>posters, slides, videos, social media, project pages, and interactive agents</b>. The core bottleneck is <b>trust</b>: researchers may use AI to draft public-facing materials, but remain reluctant to delegate final communication to systems that may distort results, overstate claims, or omit important limitations.</p>
<div class="stage-detail-section">
<div class="stage-detail-section-label">Sub-topics</div>
<div class="stage-detail-pills">
<span class="stage-detail-pill">Paper→Poster</span>
<span class="stage-detail-pill">Paper→Slides</span>
<span class="stage-detail-pill">Paper→Video</span>
<span class="stage-detail-pill">Paper→Social</span>
<span class="stage-detail-pill">Paper→Web</span>
<span class="stage-detail-pill">Paper→Agent</span>
</div>
</div>
<div class="stage-detail-section">
<div class="stage-detail-section-label">Representative methods</div>
<div class="stage-detail-methods">
<span class="stage-detail-method">Paper2Poster</span>
<span class="stage-detail-method">PPTAgent</span>
<span class="stage-detail-method">DeepPresenter</span>
<span class="stage-detail-method">PresentAgent</span>
<span class="stage-detail-method">Paper2Video</span>
<span class="stage-detail-method">Paper2Web</span>
<span class="stage-detail-method">Paper2Agent</span>
</div>
</div>
</div>
</article>
<article class="stage-card" data-stage="s8-1" onclick="toggleStage(this)">
<div class="stage-image"><img src="assets/auto-research/figures/stages/paper2video.png" alt="S8.1: Paper2Video"></div>
<div class="stage-body">
<div class="stage-tag">Stage 8.1</div>
<h4 class="stage-name">Paper2Video</h4>
<span class="stage-more">Know more</span>
</div>
<div class="stage-detail-content">
<p class="stage-detail-desc">Generating <b>explanatory videos</b> from papers by synchronizing script, narration, animations, and subtitles across multiple output channels. Video is the most demanding Paper2X format — it must coordinate visual, auditory, and temporal channels simultaneously while remaining <b>faithful to the paper's claims</b>.</p>
<div class="stage-detail-section">
<div class="stage-detail-section-label">Sub-topics</div>
<div class="stage-detail-pills">
<span class="stage-detail-pill">Script generation</span>
<span class="stage-detail-pill">Narration synthesis</span>
<span class="stage-detail-pill">Slide animation</span>
<span class="stage-detail-pill">Subtitle alignment</span>
<span class="stage-detail-pill">Talking-head video</span>
<span class="stage-detail-pill">Visual storytelling</span>
</div>
</div>
<div class="stage-detail-section">
<div class="stage-detail-section-label">Representative methods</div>
<div class="stage-detail-methods">
<span class="stage-detail-method">Paper2Video</span>
<span class="stage-detail-method">Preacher</span>
<span class="stage-detail-method">AutoPresent</span>
<span class="stage-detail-method">ScholarCast</span>
</div>
</div>
</div>
</article>
<article class="stage-card" data-stage="s8-2" onclick="toggleStage(this)">
<div class="stage-image"><img src="assets/auto-research/figures/stages/paper2slides.png" alt="S8.2: Paper2Slides"></div>
<div class="stage-body">
<div class="stage-tag">Stage 8.2</div>
<h4 class="stage-name">Paper2Slides</h4>
<span class="stage-more">Know more</span>
</div>
<div class="stage-detail-content">
<p class="stage-detail-desc">Automatically distilling paper content into <b>structured presentation decks</b>. Systems must extract key contributions, arrange them into a logical slide flow, and generate visual representations — bridging the gap between dense manuscript prose and <b>audience-ready slides</b>.</p>
<div class="stage-detail-section">
<div class="stage-detail-section-label">Sub-topics</div>
<div class="stage-detail-pills">
<span class="stage-detail-pill">Content extraction</span>
<span class="stage-detail-pill">Slide structuring</span>
<span class="stage-detail-pill">Layout generation</span>
<span class="stage-detail-pill">Figure repurposing</span>
<span class="stage-detail-pill">Key message distillation</span>
<span class="stage-detail-pill">Design automation</span>
</div>
</div>
<div class="stage-detail-section">
<div class="stage-detail-section-label">Representative methods</div>
<div class="stage-detail-methods">
<span class="stage-detail-method">PPTAgent</span>
<span class="stage-detail-method">DeepPresenter</span>
<span class="stage-detail-method">PresentAgent</span>
<span class="stage-detail-method">Paper2Slides</span>
<span class="stage-detail-method">SlideCrafter</span>
<span class="stage-detail-method">D2S</span>
</div>
</div>
</div>
</article>
<article class="stage-card" data-stage="s8-3" onclick="toggleStage(this)">
<div class="stage-image"><img src="assets/auto-research/figures/stages/paper2poster.png" alt="S8.3: Paper2Poster"></div>
<div class="stage-body">
<div class="stage-tag">Stage 8.3</div>
<h4 class="stage-name">Paper2Poster</h4>
<span class="stage-more">Know more</span>
</div>
<div class="stage-detail-content">
<p class="stage-detail-desc">Condensing papers into <b>visually compelling research posters</b>, requiring simultaneous summarization and visual design. The core challenge is balancing information density, spatial layout, and aesthetic coherence within a <b>single high-resolution canvas</b>.</p>
<div class="stage-detail-section">
<div class="stage-detail-section-label">Sub-topics</div>
<div class="stage-detail-pills">
<span class="stage-detail-pill">Layout planning</span>
<span class="stage-detail-pill">Content prioritization</span>
<span class="stage-detail-pill">Figure adaptation</span>
<span class="stage-detail-pill">Typography</span>
<span class="stage-detail-pill">Visual hierarchy</span>
<span class="stage-detail-pill">Template generation</span>
</div>
</div>
<div class="stage-detail-section">
<div class="stage-detail-section-label">Representative methods</div>
<div class="stage-detail-methods">
<span class="stage-detail-method">Paper2Poster</span>
<span class="stage-detail-method">PosterLLaVA</span>
<span class="stage-detail-method">AutoPoster</span>
<span class="stage-detail-method">PosterAgent</span>
</div>
</div>
</div>
</article>
</div>
<div class="stage-detail-panel"><div class="stage-detail-inner"></div></div>
<div class="stage-card-grid" style="margin-top:0.9rem;">
<article class="stage-card" data-stage="s8-4" onclick="toggleStage(this)">
<div class="stage-image"><img src="assets/auto-research/figures/stages/paper2social.png" alt="S8.4: Paper2Social"></div>
<div class="stage-body">
<div class="stage-tag">Stage 8.4</div>
<h4 class="stage-name">Paper2Social</h4>
<span class="stage-more">Know more</span>
</div>
<div class="stage-detail-content">
<p class="stage-detail-desc">Crafting <b>platform-adapted social media posts</b> from academic papers. Each platform (X, LinkedIn, Reddit, Mastodon) has distinct tone, length, and audience expectations — requiring format-aware generation that simplifies results <b>without distorting them</b>.</p>
<div class="stage-detail-section">
<div class="stage-detail-section-label">Sub-topics</div>
<div class="stage-detail-pills">
<span class="stage-detail-pill">Audience adaptation</span>
<span class="stage-detail-pill">Tone calibration</span>
<span class="stage-detail-pill">Thread generation</span>
<span class="stage-detail-pill">Platform formatting</span>
<span class="stage-detail-pill">Claim simplification</span>
<span class="stage-detail-pill">Hashtag generation</span>
</div>
</div>
<div class="stage-detail-section">
<div class="stage-detail-section-label">Representative methods</div>
<div class="stage-detail-methods">
<span class="stage-detail-method">Paper2Social</span>
<span class="stage-detail-method">TweetPaper</span>
<span class="stage-detail-method">PaperDigest</span>
<span class="stage-detail-method">SciComm</span>
</div>
</div>
</div>
</article>
<article class="stage-card" data-stage="s8-5" onclick="toggleStage(this)">
<div class="stage-image"><img src="assets/auto-research/figures/stages/paper2agent.png" alt="S8.5: Paper2Agent"></div>
<div class="stage-body">
<div class="stage-tag">Stage 8.5</div>
<h4 class="stage-name">Paper2Agent</h4>
<span class="stage-more">Know more</span>
</div>
<div class="stage-detail-content">
<p class="stage-detail-desc">Building <b>interactive agents</b> that allow readers to query a paper's methodology, results, and implications through conversation. These systems ground responses in specific paper content, enabling exploration <b>beyond the static manuscript</b>.</p>
<div class="stage-detail-section">
<div class="stage-detail-section-label">Sub-topics</div>
<div class="stage-detail-pills">
<span class="stage-detail-pill">Paper grounding</span>
<span class="stage-detail-pill">Conversational QA</span>
<span class="stage-detail-pill">MCP integration</span>
<span class="stage-detail-pill">Citation tracing</span>
<span class="stage-detail-pill">Demo interfaces</span>
<span class="stage-detail-pill">Fidelity evaluation</span>
</div>
</div>
<div class="stage-detail-section">
<div class="stage-detail-section-label">Representative methods</div>
<div class="stage-detail-methods">
<span class="stage-detail-method">Paper2Agent</span>
<span class="stage-detail-method">PaperCopilot</span>
<span class="stage-detail-method">DocAgent</span>
<span class="stage-detail-method">ScholarBot</span>
</div>
</div>
</div>
</article>
</div>
<div class="stage-detail-panel"><div class="stage-detail-inner"></div></div>
</div>
</div>
<script>
function toggleStage(card) {
var grid = card.closest('.stage-card-grid');
var panel = grid.nextElementSibling; // .stage-detail-panel immediately after this grid
var inner = panel.querySelector('.stage-detail-inner');
// Click on already-active card -> close
if (card.classList.contains('is-active')) {
panel.classList.remove('is-open');
card.classList.remove('is-active');
return;
}
// Close all panels + deactivate all cards in this phase-block
var phase = card.closest('.phase-block');
phase.querySelectorAll('.stage-detail-panel.is-open').forEach(function(p) {
p.classList.remove('is-open');
});
phase.querySelectorAll('.stage-card.is-active').forEach(function(c) {
c.classList.remove('is-active');
});
// Activate this card
card.classList.add('is-active');
// Render detail content from card's hidden block
var content = card.querySelector('.stage-detail-content');
var stageNum = card.querySelector('.stage-tag').textContent;
var stageName = card.querySelector('.stage-name').innerHTML;
inner.innerHTML =
'<button class="stage-detail-close" onclick="closeStage(this)" aria-label="Close">×</button>' +
'<div class="stage-detail-header">' +
'<span class="stage-detail-num">' + stageNum + '</span>' +
'<span class="stage-detail-title">' + stageName + '</span>' +
'</div>' +
content.innerHTML;
panel.classList.add('is-open');
// Smooth scroll to keep panel in view
setTimeout(function() {
var rect = panel.getBoundingClientRect();
if (rect.bottom > window.innerHeight) {
panel.scrollIntoView({ behavior: 'smooth', block: 'nearest' });
}
}, 200);
}
function closeStage(btn) {
var panel = btn.closest('.stage-detail-panel');
panel.classList.remove('is-open');
var phase = panel.closest('.phase-block');
if (phase) {
phase.querySelectorAll('.stage-card.is-active').forEach(function(c) {
c.classList.remove('is-active');
});
}
}
</script>
<br><br>
</div>
</section>
<!-- ==================== Paper Collection ==================== -->
<section id="collection">
<div class="wb-section">
<h2>Paper Collection</h2>
<div class="wb-kicker">
Per-Stage Inventory of Surveyed Methods
</div>
<div class="ar-hr"></div>
<p class="ar-lead">
A curated, expandable inventory of method discussed in the survey, organized by stage. Click a stage to expand its full table; use the search box to filter by method name, venue, or category.
</p>
<!-- Search bar -->
<label class="collection-search">
<i class="fas fa-search"></i>
<input type="text" id="collection-search-input" placeholder="Search methods, venues, or categories…" oninput="filterCollection(this.value)">
</label>
<!-- Phase filter -->
<div class="collection-filter">
<button class="cf-btn is-active" onclick="filterPhase(this,'all')">All Stages</button>
<button class="cf-btn cf-p1" onclick="filterPhase(this,'p1')"><span class="cf-dot"></span>P1 · Creation</button>
<button class="cf-btn cf-p2" onclick="filterPhase(this,'p2')"><span class="cf-dot"></span>P2 · Writing</button>
<button class="cf-btn cf-p3" onclick="filterPhase(this,'p3')"><span class="cf-dot"></span>P3 · Validation</button>
<button class="cf-btn cf-p4" onclick="filterPhase(this,'p4')"><span class="cf-dot"></span>P4 · Dissemination</button>
</div>
<div class="collection-stages">
<!-- ========== Stage 1: Idea Generation ========== -->
<div class="cstage s1" id="cstage-s1">
<div class="cstage-header" onclick="toggleCStage(this)">
<img class="cstage-thumb" src="assets/auto-research/figures/stages/s1.png" alt="">
<span class="cstage-num">S1</span>
<span class="cstage-title">Idea Generation</span>
<span class="cstage-meta">30 methods</span>
<span class="cstage-expand-btn">Expand</span>
<i class="cstage-chevron fas fa-chevron-down"></i>
</div>
<div class="cstage-body">
<div class="cstage-body-inner">
<table class="ctable">
<thead>
<tr>
<th class="col-num">#</th>
<th class="col-method">Method</th>
<th class="col-venue">Venue</th>
<th class="col-cat">Category</th>
<th class="col-note">Evaluation</th>
<th class="col-links">Links</th>
</tr>
</thead>
<tbody>
<tr class="ct-group ct-g1"><td colspan="6">LLM Internal Knowledge-Based Generation</td></tr>
<tr>
<td class="col-num">1</td>
<td class="col-method"><a href="https://arxiv.org/abs/2410.13185" target="_blank">Chain of Ideas</a></td>
<td class="col-venue"><span class="venue-pill v-conf">ACL'24</span></td>
<td class="col-cat"><span class="cat-text">LLM Internal</span></td>
<td class="col-note">Comparable to human quality; ~$0.50/idea minimum cost.</td>
<td class="col-links">
<a href="https://arxiv.org/abs/2410.13185" target="_blank" aria-label="Paper"><i class="fas fa-external-link-alt"></i></a>
<a href="https://github.com/DAMO-NLP-SG/CoI-Agent" target="_blank" aria-label="GitHub"><i class="fab fa-github"></i></a>
</td>
</tr>
<tr>
<td class="col-num">2</td>
<td class="col-method"><a href="https://aclanthology.org/2025.naacl-long.342/" target="_blank">ResearchAgent</a></td>
<td class="col-venue"><span class="venue-pill v-conf">NAACL'25</span></td>
<td class="col-cat"><span class="cat-text">LLM Internal</span></td>
<td class="col-note">Human + model evaluation; academic graph feedback for refinement.</td>
<td class="col-links">
<a href="https://aclanthology.org/2025.naacl-long.342/" target="_blank" aria-label="Paper"><i class="fas fa-external-link-alt"></i></a>
<a href="https://github.com/JinheonBaek/ResearchAgent" target="_blank" aria-label="GitHub"><i class="fab fa-github"></i></a>
</td>
</tr>
<tr>
<td class="col-num">3</td>
<td class="col-method"><a href="https://arxiv.org/abs/2305.14259" target="_blank">SciMON</a></td>
<td class="col-venue"><span class="venue-pill v-conf">ACL'24</span></td>
<td class="col-cat"><span class="cat-text">LLM Internal</span></td>
<td class="col-note">Mitigates shallow novelty via iterative refinement.</td>
<td class="col-links">
<a href="https://arxiv.org/abs/2305.14259" target="_blank" aria-label="Paper"><i class="fas fa-external-link-alt"></i></a>
<a href="https://github.com/EagleW/CLBD" target="_blank" aria-label="GitHub"><i class="fab fa-github"></i></a>
</td>
</tr>
<tr>
<td class="col-num">4</td>
<td class="col-method"><a href="https://arxiv.org/abs/2409.04109" target="_blank">Idea Gen Agent</a></td>
<td class="col-venue"><span class="venue-pill v-arxiv">arXiv'24</span></td>
<td class="col-cat"><span class="cat-text">LLM Internal</span></td>
<td class="col-note">100+ NLP researchers; LLM ideas higher novelty (p<0.05).</td>
<td class="col-links">
<a href="https://arxiv.org/abs/2409.04109" target="_blank" aria-label="Paper"><i class="fas fa-external-link-alt"></i></a>
</td>
</tr>
<tr>
<td class="col-num">5</td>
<td class="col-method"><a href="https://aclanthology.org/2025.acl-demo.57/" target="_blank">IRIS</a></td>
<td class="col-venue"><span class="venue-pill v-conf">ACL'25</span></td>
<td class="col-cat"><span class="cat-text">LLM Internal</span></td>
<td class="col-note">MCTS adaptive reasoning; human-in-the-loop platform.</td>
<td class="col-links">
<a href="https://aclanthology.org/2025.acl-demo.57/" target="_blank" aria-label="Paper"><i class="fas fa-external-link-alt"></i></a>
<a href="https://github.com/Anikethh/IRIS-Interactive-Research-Ideation-System" target="_blank" aria-label="GitHub"><i class="fab fa-github"></i></a>
</td>
</tr>
<tr>
<td class="col-num">6</td>
<td class="col-method"><a href="https://arxiv.org/abs/2504.20090" target="_blank">Spark</a></td>
<td class="col-venue"><span class="venue-pill v-conf">ICCC'25</span></td>
<td class="col-cat"><span class="cat-text">LLM Internal</span></td>
<td class="col-note">Judge model trained on 600K OpenReview reviews.</td>
<td class="col-links">
<a href="https://arxiv.org/abs/2504.20090" target="_blank" aria-label="Paper"><i class="fas fa-external-link-alt"></i></a>
</td>
</tr>
<tr>
<td class="col-num">7</td>
<td class="col-method"><a href="https://arxiv.org/abs/2512.23707" target="_blank">Rubric Rewards</a></td>
<td class="col-venue"><span class="venue-pill v-arxiv">arXiv'25</span></td>
<td class="col-cat"><span class="cat-text">LLM Internal</span></td>
<td class="col-note">70% expert preference; RL with rubric self-grading.</td>
<td class="col-links">
<a href="https://arxiv.org/abs/2512.23707" target="_blank" aria-label="Paper"><i class="fas fa-external-link-alt"></i></a>
</td>
</tr>
<tr>
<td class="col-num">8</td>
<td class="col-method"><a href="https://arxiv.org/abs/2602.18920" target="_blank">DeepInnovator</a></td>
<td class="col-venue"><span class="venue-pill v-arxiv">arXiv'26</span></td>
<td class="col-cat"><span class="cat-text">LLM Internal</span></td>
<td class="col-note">80–94% win rates vs. frontier models; 14B parameters.</td>
<td class="col-links">
<a href="https://arxiv.org/abs/2602.18920" target="_blank" aria-label="Paper"><i class="fas fa-external-link-alt"></i></a>
<a href="https://github.com/HKUDS/DeepInnovator" target="_blank" aria-label="GitHub"><i class="fab fa-github"></i></a>
</td>
</tr>
<tr class="ct-group ct-g2"><td colspan="6">External Signal-Driven Generation</td></tr>
<tr>
<td class="col-num">9</td>
<td class="col-method"><a href="https://openreview.net/forum?id=X9OfMNNepI" target="_blank">MOOSE-Chem</a></td>
<td class="col-venue"><span class="venue-pill v-top">ICLR'25</span></td>
<td class="col-cat"><span class="cat-text">External Signal</span></td>
<td class="col-note">Rediscovers hypotheses from 51 high-impact papers.</td>
<td class="col-links">
<a href="https://openreview.net/forum?id=X9OfMNNepI" target="_blank" aria-label="Paper"><i class="fas fa-external-link-alt"></i></a>
</td>
</tr>
<tr>
<td class="col-num">10</td>
<td class="col-method"><a href="https://arxiv.org/abs/2410.14255" target="_blank">Nova</a></td>
<td class="col-venue"><span class="venue-pill v-arxiv">arXiv'24</span></td>
<td class="col-cat"><span class="cat-text">External Signal</span></td>
<td class="col-note">3.4× more novel ideas; 2.5× more top-rated.</td>
<td class="col-links">
<a href="https://arxiv.org/abs/2410.14255" target="_blank" aria-label="Paper"><i class="fas fa-external-link-alt"></i></a>
</td>
</tr>
<tr>
<td class="col-num">11</td>
<td class="col-method"><a href="https://arxiv.org/abs/2409.05556" target="_blank">SciAgents</a></td>
<td class="col-venue"><span class="venue-pill v-arxiv">arXiv'24</span></td>
<td class="col-cat"><span class="cat-text">External Signal</span></td>
<td class="col-note">Multi-agent reasoning over scientific knowledge graphs.</td>
<td class="col-links">
<a href="https://arxiv.org/abs/2409.05556" target="_blank" aria-label="Paper"><i class="fas fa-external-link-alt"></i></a>
<a href="https://github.com/lamm-mit/SciAgentsDiscovery" target="_blank" aria-label="GitHub"><i class="fab fa-github"></i></a>
</td>
</tr>
<tr>
<td class="col-num">12</td>
<td class="col-method"><a href="https://arxiv.org/abs/2410.23166" target="_blank">SciPIP</a></td>
<td class="col-venue"><span class="venue-pill v-arxiv">arXiv'24</span></td>
<td class="col-cat"><span class="cat-text">External Signal</span></td>
<td class="col-note">Multi-domain; paper-anchored idea generation.</td>
<td class="col-links">
<a href="https://arxiv.org/abs/2410.23166" target="_blank" aria-label="Paper"><i class="fas fa-external-link-alt"></i></a>
<a href="https://github.com/cheerss/SciPIP" target="_blank" aria-label="GitHub"><i class="fab fa-github"></i></a>
</td>
</tr>
<tr>
<td class="col-num">13</td>
<td class="col-method"><a href="https://arxiv.org/abs/2410.04025" target="_blank">IdeaSynth</a></td>
<td class="col-venue"><span class="venue-pill v-conf">CHI'25</span></td>
<td class="col-cat"><span class="cat-text">External Signal</span></td>
<td class="col-note">20-user study; more alternatives explored vs. baseline.</td>
<td class="col-links">
<a href="https://arxiv.org/abs/2410.04025" target="_blank" aria-label="Paper"><i class="fas fa-external-link-alt"></i></a>
</td>
</tr>
<tr>
<td class="col-num">14</td>
<td class="col-method"><a href="https://nips.cc/virtual/2025/poster/118171" target="_blank">MOOSE-Chem2</a></td>
<td class="col-venue"><span class="venue-pill v-top">NeurIPS'25</span></td>
<td class="col-cat"><span class="cat-text">External Signal</span></td>
<td class="col-note">Fine-grained, experimentally actionable hypotheses.</td>
<td class="col-links">
<a href="https://nips.cc/virtual/2025/poster/118171" target="_blank" aria-label="Paper"><i class="fas fa-external-link-alt"></i></a>
</td>
</tr>
<tr>
<td class="col-num">15</td>
<td class="col-method"><a href="https://arxiv.org/abs/2603.29557" target="_blank">FlowPIE</a></td>
<td class="col-venue"><span class="venue-pill v-arxiv">arXiv'26</span></td>
<td class="col-cat"><span class="cat-text">External Signal</span></td>
<td class="col-note">Higher novelty, feasibility, and diversity vs. baselines.</td>
<td class="col-links">
<a href="https://arxiv.org/abs/2603.29557" target="_blank" aria-label="Paper"><i class="fas fa-external-link-alt"></i></a>
</td>
</tr>
<tr class="ct-group ct-g3"><td colspan="6">Multi-Agent Collaborative Generation</td></tr>
<tr>
<td class="col-num">16</td>
<td class="col-method"><a href="https://arxiv.org/abs/2412.14141" target="_blank">Combi. Creativity</a></td>
<td class="col-venue"><span class="venue-pill v-arxiv">arXiv'24</span></td>
<td class="col-cat"><span class="cat-text">Multi-Agent</span></td>
<td class="col-note">+7–10% similarity scores; cross-domain composition.</td>
<td class="col-links">
<a href="https://arxiv.org/abs/2412.14141" target="_blank" aria-label="Paper"><i class="fas fa-external-link-alt"></i></a>