Tabularis Formatus: Predictive Formatting for Tables

Gust Verbruggen; Jos\'e Cambronero; Mukul Singh; Sumit Gulwani; Vu Le

REVIEW 3 major objections 7 minor 55 references

Tabularis Formatus: Predictive Formatting for Tables

T0 review · 3 major / 7 minor · reviewed 2026-08-15 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Spreadsheet conditional-formatting rules can be predicted from the table alone, with no user examples or instructions.

desk verdict New task and a well-engineered neuro-symbolic pipeline, but possible format-retrieval leakage and a suspiciously high end-to-end number need resolution before the headline claims hold. read the letter →

arxiv 2508.11121 v1 pith:GNZ6JLZN submitted 2025-08-14 cs.DB cs.AIcs.SE

Mukul Singh , José Cambronero , Sumit Gulwani , Vu Le , Gust Verbruggen This is my paper · ORCID

classification cs.DBcs.AIcs.SE

keywords conditionalformattingpredictivespreadsheetautomationprogramsynthesisneuro-symbolicsystemslargelanguagemodelsexecution-guidedrankingtable

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper sets out to show that spreadsheet conditional formatting can be made fully predictive: given a table and a target column, a system called Tafo suggests both the condition that triggers a style change and the visual format to apply, without the user supplying examples, natural-language instructions, or any other intent. The authors argue this matters because most users either do not know which rules to create or find the rule interface difficult, and prior automation required some form of specification. On a benchmark of 105K conditional-formatting tasks and 100K manually formatted columns drawn from 1.8 million public workbooks, Tafo matches the user's own formatting on over half of the tasks with just three suggestions. It also reports execution-match accuracy of 36.7%, 46.8%, and 64.3% at top-1, top-3, and top-5 suggestions, with 15.6 to 26.5 percentage points higher execution-match accuracy than the adapted neural and symbolic baselines.

What carries the argument

The mechanism that carries the argument is a three-way neuro-symbolic generator followed by execution-based ranking. The symbolic arm performs component-based synthesis over typed static properties, including counts of blanks, errors, and formulas, most common values, statistical summaries, and text and date predicates, with beam search driven by a learned ranker. The neural arm prompts a large language model in a fixed four-step chain of thought: pick relevant columns, propose predicates and functions, list constants, then write rules; Tafo parses these generations into predicate components that re-enter the symbolic enumerator with an extra 10% score boost. The ranking step executes every candidate rule on the table, clusters rules whose boolean output vectors are identical into execution-equivalence classes, scores each cluster by the average member score, and samples round-robin across clusters, so the final suggestions differ in the cells they highlight. The format learner is a separate corpus-retrieval step that finds similar table-rule pairs by comparing headers, formulas, and rule structure with a weighted Hamming distance, mines their format identifiers, and grounds the chosen color to the shade used by at least 75% of the sheet's formatted cells.

What would settle it

Take a fresh sample of spreadsheets with conditional-formatting rules, ask the users who wrote them to rate whether Tafo's top-three suggestions capture what they intended, and compare that acceptance rate with execution match; if accepted suggestions are far rarer than execution match, the benchmark's ground truth overstates practical automation.

Watch

Extended reading notes

Core claim

The central discovery, on the paper's own terms, is that the two halves of a conditional-formatting rule, the trigger and the format, can be learned together from the data itself. Tafo does this by pooling candidates from three generators: a symbolic enumerator that builds predicates from typed static properties, a large-language-model generator that reasons about the table in four steps, and a neuro-symbolic generator that feeds the LLM's outputs back into the symbolic search with boosted weight. All candidates are executed on the table and clustered by the set of cells they highlight; clusters are scored and sampled round-robin, which is what makes the final shortlist both accurate and diverse. The format half is learned by retrieving similar table-rule pairs from a large corpus, mining the format identifiers used there, weighting formats already present in the current sheet double, and grounding colors to the sheet's dominant shade. Evaluated on 105K conditional-formatting tasks, Tafo reaches 64.3% execution match at top-5, 46.8% at top-3, and about 50% end-to-end condition-plus-format match with three suggestions.

Load-bearing premise

The accuracy numbers treat the formatting users actually applied in 1.8 million public workbooks, their conditional-formatting rules and manually highlighted cells, as the correct target, so the claim assumes this real-world formatting encodes shared, desirable semantics rather than idiosyncratic taste, habit, or mistakes.

Editorial extensions

If this is right

If the accuracy numbers hold, spreadsheet users can receive useful conditional-formatting suggestions without writing anything: three suggestions reproduce the user's own formatting on more than half of both rule-based and manually formatted columns.
Because suggestions are ranked by execution, the shortlist is not a set of near-duplicates: two suggestions already cover over 80% of the column's cells, so a user scanning three options sees rules that highlight different parts of the data.
The two halves of formatting are separable but coupled: a rule trigger can be suggested even when the user never wrote a rule, since the same pipeline matches manually highlighted cells, and the format can be suggested even when the condition is simple.
The ablations imply the design's main levers: removing symbolic property extraction costs about 19.5 points of top-1 execution match, and removing execution-guided ranking costs about 20.4 points at top-5.
The corpus-driven format learner implies that format suggestions are not fixed templates but adapt to the table's existing style, so the same condition can be suggested with the user's shade of green rather than a default color.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

Inference: the execution-match metric makes the rule itself interchangeable with any other rule that highlights the same cells, so a production system could present highlighting patterns directly to users and only generate a symbolic rule when a user asks to see or edit it.
Inference: the same corpus-retrieval machinery that finds similar tables for formats could be turned into a recommender for other data-dependent artifacts such as charts, pivot tables, or data-validation rules, and the paper's own reliance on similar-table sharing gives a route to test that.
Inference: the shade-grounding heuristic, which uses a shade if 75% of the sheet's formatted cells use it, is a simple personalization rule; replacing it with a per-user or per-document learned color model is a direct way to test whether top-1 format match can move beyond the reported 13.2% exact match.
Inference: because the ground truth is formatting users actually applied in public workbooks, the headline automation rate is an upper bound on what users would accept; a deployment study measuring whether users keep the suggestions would be a sharper test of practical value.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Read the Tafo paper. The new thing here is real: fully predictive conditional formatting — no examples, no NL — with a neuro-symbolic pipeline that combines symbolic predicate enumeration, LLM-generated semantic hints, and execution-based diversity ranking. The evaluation is large-scale (1.8M workbooks, 25K held-out CF tasks plus 100K manual-format tasks), and the condition-learning results in Table 3 look internally consistent: Tafo beats the best neural baseline by roughly 5-7 points on execution match at top-1. The format learner that mines similar tables for candidate colors and shades is a sensible idea, and the ablation showing corpus retrieval matters is informative. Full credit for tackling a task nobody else has defined cleanly.

But there are two soft spots that matter.

First, the format learner may be reading the answer key. Section 4.4 retrieves candidate formats from similar tables and rules in the corpus using weighted Hamming distance on headers, formulas, predicates, and constants. The evaluation tasks are drawn from that same corpus. The paper never states that the target workbook or its near-duplicates are excluded from the retrieval pool. Since a table is most similar to itself, the user's own format can enter the candidate set directly. That would inflate format quality numbers and, in turn, the end-to-end 50% claim — and it would do so specifically for Tafo, since the baselines do not do corpus retrieval. This is not a minor nit; if the authors cannot rule out self-retrieval, the central comparison in Table 4 is compromised.

Second, the numbers do not quite add up. Table 3 gives Tafo top-3 execution match for conditions at 46.8%. Table 4 gives top-3 exact format match at 18.9% and property match at 47.2%. Yet Figure 11 and the text claim 50% end-to-end execution match with three suggestions. Since end-to-end is defined as condition execution match AND format color match, 50% is suspiciously high. Either the metric is defined differently (e.g., measured only over tasks where the condition matched), or there is an evaluation-set mismatch. The paper needs to clarify this.

Minor issues: no error bars or significance tests; several dangling references (Appendix??, Figure??); no artifact release mentioned; and the manual-formatting benchmark lacks a simple baseline like repeating the most common existing format in the column. The noisy-oracle concern — user formatting is not always semantically meaningful — is real but applies equally to all systems, so it is an external-validity caveat rather than a fatal flaw.

Who is this for? Researchers in spreadsheet automation and neuro-symbolic program synthesis. The task formulation is worth publishing, and the pipeline has reusable ideas, especially execution clustering and LLM-to-symbolic hints. But the paper should not be accepted until the retrieval leakage and the metric discrepancy are addressed.

Yes, it deserves a serious referee — the novelty is sufficient. My own read, though, is that the headline accuracy numbers are probably too high as stated.

Referee Report

3 major / 7 minor

Summary. The paper introduces Tafo, a neuro-symbolic system for predicting conditional formatting (CF) suggestions for spreadsheet tables without any user specification. Tafo generates rule conditions through three complementary mechanisms — a purely symbolic enumerator, an LLM-based multi-step reasoning generator, and a neuro-symbolic combination — then ranks candidates using execution-based clustering and learns visual formats by retrieving similar (table, rule) pairs from a corpus. The system is evaluated on 25K CF tasks and 100K manually formatted column tasks extracted from 1.8 million public Excel workbooks. The authors report that Tafo outperforms a range of symbolic and neural baselines by 15.6%–26.5% on execution match, achieves 50% end-to-end execution match with three suggestions, and attains higher format exact/property match and condition diversity. They also present ablations of the condition generator, the ranker, and the format learner, as well as robustness checks on sample size and data type.

Significance. The paper introduces a practically important task — fully predictive formatting suggestion — and a technically plausible architecture that combines symbolic synthesis, LLM semantic knowledge, and corpus-based format retrieval. The evaluation is broad: it compares against six symbolic and seven neural baselines, uses multiple matching metrics (exact, sketch, execution; color and property match), and includes diversity, coverage, complexity, latency, and ablation analyses. If the reported gains are reproducible after addressing the methodological concerns below, Tafo would be a meaningful advance for spreadsheet automation and for neuro-symbolic recommenders more generally. The strengths are the clear problem formalization, the systematic baseline selection, and the careful construction of the CF and manual-formatting benchmarks from a large real-world corpus.

major comments (3)

[Section 4.4 and Section 5.1] The format learner retrieves candidate (table, rule) pairs from the same 1.8M-workbook corpus that also defines the evaluation ground truth, and the paper does not state that the query workbook or its near-duplicates are excluded from the retrieval pool. Since a table is by definition the most similar item to itself, the format learner can directly copy the user's own format for that table, which would inflate the format-match numbers in Table 4 and the end-to-end numbers in Figure 11. This is load-bearing for the headline claims of 50% automation and 15.6%–26.5% relative gains. Please state explicitly whether the target workbook is removed from the retrieval set, and if not, re-run the format learner with a retrieval pool that excludes the target workbook and near-duplicates (e.g., by workbook ID or min-hash deduplication) and report the resulting Table 4 and Figure 11 numbers.
[Section 6, Tables 3–4, Figures 10–11] The headline comparisons are reported as point estimates with no confidence intervals or significance tests. For example, Table 3 shows a top-5 execution match of 64.3% for Tafo versus 58.3% for GPT-4, and Table 4 shows top-1 exact format match of 13.2% versus 9.5% for FormaT5; with 25K tasks these differences may be significant, but without noise estimates the reader cannot assess reliability. Please report the number of task-level replicates, compute bootstrap confidence intervals or paired significance tests for the main metrics, and clarify what the 'Error Bounds' in Figure 10 represent (currently undefined).
[Section 5.1 and Section 6.1.1] The benchmark treats user-applied formatting as the ground-truth target with no check for formatting that is stylistic, idiosyncratic, or erroneous. This assumptions affects the external-validity interpretation of 'automating over 50% of user formatting tasks' — a system might match arbitrary user color choices rather than semantically meaningful highlighting. I recommend reporting a human evaluation on a sample of tasks to measure whether the top-ranked suggestions are judged useful, and/or an analysis on a subset of tasks where user formatting follows common semantic patterns (e.g., green for positive, red for negative, bold for maxima). At minimum, a limitation paragraph acknowledging this issue should be added.

minor comments (7)

[Table 4 caption] The word 'accross' should be 'across'.
[Abstract and body] The system name is written as both 'TaFo' and 'Tafo'; please use one consistent spelling throughout.
[Figure 15] The y-axis label 'TabuTran' appears to be a typo; it should likely read 'Tafo'.
[Section 6.1.1 and Section 7] The manual-formatting automation numbers are inconsistent: Section 1 says 'over 50%', Figure 10 shows 55% at top-3, and Section 7 says 'over 60%'. Please reconcile these claims.
[Section 5.3 and Section 6.3] There are unresolved placeholders: 'Implementation details are in Appendix??' and 'Figure?? shows...' must be replaced with actual references.
[CCS Concepts and Keywords] The CCS Concepts and Keywords fields contain placeholder text ('Do Not Use This Code', 'Do, Not, Us, This, Code, Put, the, Correct, Terms, for, Your, Paper') and must be replaced.
[Figure 2 caption] The phrase 'After the conidition' contains a typo; it should be 'After the condition'.

Circularity Check

1 steps flagged · score 6.0 of 10

Format learner retrieves candidate formats from the same corpus that defines ground truth, so format and end-to-end gains may be partly forced by retrieval rather than generalization.

fitted input called prediction [Section 4.4 (format learner) combined with Section 5.1 (benchmark) and Table 7 (format ablations)]
"First, we collect candidate formats f_c = f_e ∪ f_r with f_e being all formats already present in the current sheet and f_r the formats from similar tables and rules in the corpus. ... We use 80K tasks for training the ranker and rule generation engine, and the remaining 25K tasks for evaluation."

The format learner's output is selected from f_r, formats mined from similar tables and rules in the same 1.8M-workbook corpus that supplies the benchmark's ground-truth user formats. No statement excludes the target workbook or near-duplicates from the retrieval pool, so for a test task the ground-truth format can be in f_r by construction; reporting a match against it then measures retrieval of the answer key, not generalization. Table 7 confirms corpus retrieval is the dominant format component (-19.1 to -9.0 points), and Figure 11's end-to-end 50% claim depends on this format score, making the headline partially forced.

full rationale

Tafo's condition learning is evaluated on a held-out 25K-task split with a ranker trained on 80K tasks, which is standard supervised evaluation and not circular. The LLM and symbolic generators are also not circular. Self-citations to Cornet and FormaT5 are earlier work used as baselines and data source; they are not load-bearing as unverified authorities. The material circularity is in the format learner: it retrieves candidate formats from the corpus, and the benchmark's ground-truth formats are user formats in that same corpus. The paper never states that the evaluation workbooks are removed from the retrieval pool. Since the target table and rule are the closest match to themselves, the ground-truth format can enter f_r directly, so format accuracy and the end-to-end number are partly a retrieval artifact. This is a partial reduction of the central claim, not a full one, because the condition component is held out and the format weights come from 100 annotated pairs. Score 6.

Assumptions & free parameters 7 free parameters · 5 assumptions · 0 invented entities

The system is evaluated empirically and every component rests on the corpus of 1.8M public workbooks introduced in the authors' prior Cornet paper. The format learner in particular is fitted to human annotations (100 pairs) and has several hand-set thresholds; the ground-truth validity is assumed, not shown. No new physical or conceptual entities are introduced.

free parameters (7)

Corpus retrieval cutoffs lambda_N and lambda_T = not reported
Section 4.4 stops mining similar rule-table pairs until lambda_N pairs or similarity below lambda_T; values not given, tuned on validation.
Similarity weights for table/rule matching = from linear regression on 100 manually annotated pairs
Section 4.4: weights are obtained by manually annotating the similarity of 100 tables and performing linear regression over these; fitted to human judgments, not an external benchmark.
Current-sheet format weighting factor = 2x
Section 4.4: formats in f_e have twice the weight as those in f_r; hand-set.
Shade grounding threshold = 75%
Section 4.4: if at least 75% of cells has a specific shade, we suggest that shade; hand-set threshold.
Neural component reward = 10% of node score
Section 4.1: the reward is computed as 10% of the original score of the node; hand-set.
Manual-formatting column filter = >5 and <|E_j| formatted cells
Section 5.1 defines a formatted column; this filter shapes the benchmark and affects the 55% automation claim.
Ranker parameters = trained on 80K tasks, not released
Section 4.1 dense network with logistic regression objective; parameters fit to the corpus, not released for independent use.

assumptions (5)

domain assumption User-applied formatting in public workbooks encodes desirable semantic intent
Section 1 and 5.1 treat user CF rules and manual formatting as ground truth; if formatting is stylistic or idiosyncratic, the benchmark and the 50%+ automation claims lose meaning.
domain assumption Similar tables have similar formatting artifacts
Section 4.4 extends the hypothesis in [53] from formulas/charts to formats; load-bearing for the format learner, tested only on the same corpus.
domain assumption Color semantics such as green positive and red negative are widely shared
Section 1 example and Section 6.1.2 qualitative analysis assume shared color semantics hold across users.
domain assumption Text, numeric, and date column types and the chosen format identifiers cover the task
Section 3 restricts the type and format vocabularies following [40,41]; the system cannot suggest other types or formats.
ad hoc to paper Column properties and predicate templates are sufficient building blocks for rule synthesis
Section 4.1 uses a set of templates curated specific to formatting tasks, not derived from a completeness argument.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Tabularis Formatus: Predictive Formatting for Tables." pith.science (2026). https://pith.science/paper/GNZ6JLZN

@misc{pith2026250811121,
  author       = {Pith},
  title        = {Pith review of: Tabularis Formatus: Predictive Formatting for Tables},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GNZ6JLZN}},
  note         = {Machine review of arXiv:2508.11121}
}

read the original abstract

Spreadsheet manipulation software are widely used for data management and analysis of tabular data, yet the creation of conditional formatting (CF) rules remains a complex task requiring technical knowledge and experience with specific platforms. In this paper we present TaFo, a neuro-symbolic approach to generating CF suggestions for tables, addressing common challenges such as user unawareness, difficulty in rule creation, and inadequate user interfaces. TaFo takes inspiration from component based synthesis systems and extends them with semantic knowledge of language models and a diversity preserving rule ranking.Unlike previous methods focused on structural formatting, TaFo uniquely incorporates value-based formatting, automatically learning both the rule trigger and the associated visual formatting properties for CF rules. By removing the dependency on user specification used by existing techniques in the form of formatted examples or natural language instruction, TaFo makes formatting completely predictive and automated for the user. To evaluate TaFo, we use a corpus of 1.8 Million public workbooks with CF and manual formatting. We compare TaFo against a diverse set of symbolic and neural systems designed for or adapted for the task of table formatting. Our results show that TaFo generates more accurate, diverse and complete formatting suggestions than current systems and outperforms these by 15.6\%--26.5\% on matching user added ground truth rules in tables.

Figures

Figures reproduced from arXiv: 2508.11121 by the authors.

**Figure 2.** The following five sections respectively describe the three [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗

**Figure 2.** Summary of Tafo on a user table (1). First, Tafo learns multiple conditions for the tables (2) by pooling candidate conditions from multiple generators (3) and ranking them using an execution based ranker (4). After the conidition, Tafo generates the associated format for the learned conditionals (5) generating the final suggestion (6) [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

Figures from the paper (12 more)

**Figure 3.** Figure 3: Symbolic rule synthesis overview. The input table [PITH_FULL_IMAGE:figures/full_fig_p004_3.png]

**Figure 5.** Figure 5: Prompt structure used to generate neural candidates [PITH_FULL_IMAGE:figures/full_fig_p005_5.png]

**Figure 6.** Figure 6: Figure summarizing the neuro-symbolic condition generator in [PITH_FULL_IMAGE:figures/full_fig_p006_6.png]

**Figure 7.** Figure 7: Suggestion ranking system. The learner generates [PITH_FULL_IMAGE:figures/full_fig_p006_7.png]

**Figure 8.** Figure 8: Summary of the format learning for Tafo. Given a user table T (1) and the learned rule triggers 𝑡 ∈ (𝑡, 𝑓 ) = 𝑟 ∈ 𝑅, Tafo retrieves similar (𝑡𝑎𝑏𝑙𝑒, 𝑟𝑢𝑙𝑒) pairs from the corpus (3) and mines formats used in them. The mined formats are then transformed based on the forma…

**Figure 9.** Figure 9: Figure showing the format grounding for a sample [PITH_FULL_IMAGE:figures/full_fig_p007_9.png]

**Figure 10.** Figure 10: Tafo performance on tasks with manual formatting. We show execution match accuracy for increasing number of samples generated (top-𝑘). 1 2 3 4 5 6 7 8 9 10 Number of Suggestions (Top-k) 20 30 40 50 60 End to End Execution Match TaFo GPT4 FormaT5 [PITH_FULL_IMAGE:fi…

**Figure 11.** Figure 11: End to end execution match for Tafo, the best baseline (GPT-4) and the current state-of-the-art in table formatting (FormaT5) against increasing number of suggestions (Top-𝑘). Tafo outperforms both baselines [PITH_FULL_IMAGE:figures/full_fig_p009_11.png]

**Figure 12.** Figure 12: (a) Completeness: Percentage tasks with greater [PITH_FULL_IMAGE:figures/full_fig_p010_12.png]

**Figure 14.** Figure 14: Execution match for condition learning for [PITH_FULL_IMAGE:figures/full_fig_p011_14.png]

**Figure 16.** Figure 16: Execution match for condition learning for [PITH_FULL_IMAGE:figures/full_fig_p012_16.png]

**Figure 15.** Figure 15: Top-3 suggestion execution match for condition [PITH_FULL_IMAGE:figures/full_fig_p012_15.png]

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

55 extracted references · 26 canonical work pages

[1]

Sergei Abramovich, Stephen Sugden, Sergei Abramovich, and Stephen J Sugden

work page
[2]

Sibei Chen, Yeye He, Weiwei Cui, Ju Fan, Song Ge, Haidong Zhang, Dong- mei Zhang, and Surajit Chaudhuri. 2024. Auto-Formula: Recommend For- mulas in Spreadsheets using Contrastive Learning for Table Representations. arXiv:2404.12608 [cs.DB]

work page Pith review arXiv 2024
[3]

Wenhu Chen, Xueguang Ma, Xinyi Wang, and William W. Cohen. 2023. Pro- gram of Thoughts Prompting: Disentangling Computation from Reasoning for Numerical Reasoning Tasks. Transactions on Machine Learning Research (2023). https://openreview.net/forum?id=YfZ4ZPt8zd

2023
[4]

Xinyun Chen, Petros Maniatis, Rishabh Singh, Charles Sutton, Hanjun Dai, Max Lin, and Denny Zhou. 2021. SpreadsheetCoder: Formula Prediction from Semi- structured Context. In Proceedings of the 38th International Conference on Machine Learning (Proceedings of Machine Learning Research, Vol. 139) , Marina Meila and Tong Zhang (Eds.). PMLR, virtual, 1661–16...

work page 2021
[5]

Haoyu Dong, Jinyu Wang, Zhouyu Fu, Shi Han, and Dongmei Zhang. 2020. Neural Formatting for Spreadsheet Tables. In Proceedings of the 29th ACM International Conference on Information & Knowledge Management (Virtual Event, Ireland) (CIKM ’20). Association for Computing Machinery, New York, NY, USA, 305–314. doi:10.1145/3340531.3411943

arXiv 2020
[6]

Kevin Ellis and Sumit Gulwani. 2017. Learning to Learn Programs from Examples: Going Beyond Program Structure. In IJCAI 2017 (ijcai 2017 ed.). IJCAI 2017, Melbourne, Australia, 1638–1645. www.microsoft.com/research/publication/lea rning-learn-programs-examples-going-beyond-program-structure/

work page 2017
[7]

Anna Fariha and Alexandra Meliou. 2019. Example-Driven Query Intent Discov- ery: Abductive Reasoning Using Semantic Similarity. Proc. VLDB Endow. 12, 11 (jul 2019), 1262–1275. doi:10.14778/3342263.3342266

arXiv 2019
[8]

Anna Fariha, Ashish Tiwari, Alexandra Meliou, Arjun Radhakrishna, and Sumit Gulwani. 2021. CoCo: Interactive Exploration of Conformance Constraints for Data Understanding and Data Cleaning. In Proceedings of the 2021 Inter- national Conference on Management of Data (Virtual Event, China) (SIGMOD ’21). Association for Computing Machinery, New York, NY, USA...

arXiv 2021

Show all 55 references

[9]

Zhangyin Feng, Daya Guo, Duyu Tang, Nan Duan, Xiaocheng Feng, Ming Gong, Linjun Shou, Bing Qin, Ting Liu, Daxin Jiang, and Ming Zhou. 2020. CodeBERT: A Pre-Trained Model for Programming and Natural Languages. In EMNLP 2020. Association for Computational Linguistics, Online, 15...

2020 doi
[10]

Sumit Gulwani. 2011. Automating String Processing in Spreadsheets using Input-Output Examples. In PoPL’11, January 26-28, 2011, Austin, Texas, USA . Association for Computing Machinery, New York, NY, USA, 317–330. https: //www.microsof t.com/en- us/research/publication/automat...

2011
[11]

Jonathan Herzig, Paweł Krzysztof Nowak, Thomas Müller, Francesco Piccinno, and Julian Martin Eisenschlos. 2020. Tapas: Weakly Supervised Table Parsing via Pre-training. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long P...

2020
[12]

Sirui Hong, Yizhang Lin, Bang Liu, Bangbang Liu, Binhao Wu, Danyang Li, Jiaqi Chen, Jiayi Zhang, Jinlin Wang, Li Zhang, Lingyao Zhang, Min Yang, Mingchen Zhuge, Taicheng Guo, Tuo Zhou, Wei Tao, Wenyi Wang, Xiangru Tang, Xiang- tao Lu, Xiawu Zheng, Xinbing Liang, Yaying Fei, Yu...

2024 arXiv
[13]

Yupeng Hou, Junjie Zhang, Zihan Lin, Hongyu Lu, Ruobing Xie, Julian McAuley, and Wayne Xin Zhao. 2024. Large Language Models are Zero-Shot Rankers for Recommender Systems. arXiv:2305.08845 [cs.IR]

2024 arXiv
[14]

Nathan Hurst, Kim Marriott, and Peter Moulder. 2005. Toward tighter tables. In Proceedings of the 2005 ACM symposium on Document engineering . Association for Computing Machinery, New York, NY, USA, 74–83

2005
[15]

Harshit Joshi, Abishai Ebenezer, José Cambronero, Sumit Gulwani, Aditya Kanade, Vu Le, Ivan Radiček, and Gust Verbruggen. 2023. FLAME: A small language model for spreadsheet formulas. arXiv:2301.13779 [cs.PL]

2023 arXiv
[16]

Anirudh Khatry, Joyce Cahoon, Jordan Henkel, Shaleen Deep, Venkatesh Emani, Avrilia Floratou, Sumit Gulwani, Vu Le, Mohammad Raza, Sherry Shi, Mukul Singh, and Ashish Tiwari. 2023. From Words to Code: Harnessing Data for Program Synthesis from Natural Language. arXiv:2305.0159...

2023 arXiv
[17]

Vu Le and Sumit Gulwani. 2014. FlashExtract: a framework for data extraction by examples. In 2014 Programming Language Design and Implementation . ACM, New York, NY, USA, 542–553. https://www.microsoft.com/en-us/research/publ ication/flashextract-framework-data-extraction-examples/

2014
[18]

Hao Li, Chee-Yong Chan, and David Maier. 2015. Query from Examples: An Iterative, Data-Driven Approach to Query Construction. Proc. VLDB Endow. 8, 13 (sep 2015), 2158–2169. doi:10.14778/2831360.2831369

2015
[19]

Raymond Li, Loubna Ben Allal, Yangtian Zi, Niklas Muennighoff, Denis Kocetkov, Chenghao Mou, Marc Marone, Christopher Akiki, Jia Li, Jenny Chim, Qian Liu, Evgenii Zheltonozhskii, Terry Yue Zhuo, Thomas Wang, Olivier Dehaene, Mishig Davaadorj, Joel Lamy-Poirier, João Monteiro, ...

2023 arXiv
[20]

Pietro Liguori, Erfan Al-Hossami, Domenico Cotroneo, Roberto Natella, Bojan Cukic, and Samira Shaikh. 2022. Can we generate shellcodes via natural language? An empirical study. Automated Software Engineering 29 (2022), 1–34

2022
[21]

Xiaofan Lin. 2006. Active layout engine: Algorithms and applications in variable data printing. Computer-Aided Design 38, 5 (2006), 444–456

2006
[22]

Ziyang Luo, Can Xu, Pu Zhao, Qingfeng Sun, Xiubo Geng, Wenxiang Hu, Chongyang Tao, Jing Ma, Qingwei Lin, and Daxin Jiang. 2023. Wiz- ardCoder: Empowering Code Large Language Models with Evol-Instruct. arXiv:2306.08568 [cs.CL]

2023 arXiv
[23]

James MacQueen et al. 1967. Some methods for classification and analysis of multivariate observations. In Proceedings of the fifth Berkeley symposium on mathematical statistics and probability , Vol. 1. Oakland, CA, USA, 281–297

1967
[24]

Paula Maddigan and Teo Susnjak. 2023. Chat2VIS: Generating Data Visualisations via Natural Language using ChatGPT, Codex and GPT-3 Large Language Models. arXiv:2302.02094 [cs.HC]

2023 arXiv
[25]

Davide Mottin, Matteo Lissandrini, Yannis Velegrakis, and Themis Palpanas. 2016. Exemplar Queries: A New Way of Searching. The VLDB Journal 25, 6 (dec 2016), 741–765. doi:10.1007/s00778-016-0429-2

2016 doi
[26]

Joseph N. 2022. Number of Google Sheets and Excel Users Worldwide. https:// askwonder.com/research/number-google-sheets-users-worldwide-eoskdoxav. Last Accessed: 2022-07-30

2022
[27]

Nagarajan Natarajan, Danny Simmons, Naren Datha, Prateek Jain, and Sumit Gulwani. 2019. Learning Natural Programs from a Few Examples in Real-Time. In AIStats. PMLR, online, 1714–1722. https://www.microsoft.com/en-us/researc h/publication/learning-natural-programs-from-a-few-e...

2019
[28]

Roberto Navigli, Simone Conia, and Björn Ross. 2023. Biases in Large Language Models: Origins, Inventory, and Discussion. J. Data and Information Quality 15, 2, Article 10 (jun 2023), 21 pages. doi:10.1145/3597307

2023 doi
[29]

Erich Neuwirth and Deane Arganbright. 2003. The Active Modeler: Mathematical Modeling With Microsoft Excel . Duxbury Press, online

2003
[30]

Erik Nijkamp, Bo Pang, Hiroaki Hayashi, Lifu Tu, Huan Wang, Yingbo Zhou, Silvio Savarese, and Caiming Xiong. 2023. CodeGen: An Open Large Language Model for Code with Multi-Turn Program Synthesis. arXiv:2203.13474 [cs.LG]

2023 arXiv
[31]

OpenAI. 2024. GPT-4 Technical Report. arXiv:2303.08774 [cs.CL]

2024 arXiv
[32]

Ari Pakman, Yueqi Wang, Catalin Mitelut, JinHyung Lee, and Liam Paninski
[33]

Justin Payan, Swaroop Mishra, Mukul Singh, Carina Negreanu, Christian Poelitz, Chitta Baral, Subhro Roy, Rasika Chakravarthy, Benjamin Van Durme, and Elnaz Nouri. 2023. InstructExcel: A Benchmark for Natural Language Instruction in Excel. arXiv:2310.14495 [cs.CL] https://arxiv...

2023 arXiv
[34]

Gabriel Poesia, Oleksandr Polozov, Vu Le, Ashish Tiwari, Gustavo Soares, Christo- pher Meek, and Sumit Gulwani. 2022. Synchromesh: Reliable code generation from pre-trained language models. CoRR abs/2201.11227 (2022). arXiv:2201.11227 https://arxiv.org/abs/2201.11227

2022 arXiv
[35]

Copet, Faisal Azhar, Hugo Touvron, Louis Martin, Nicolas Usunier, Thomas Scialom, and Gabriel Synnaeve

Baptiste Rozière, Jonas Gehring, Fabian Gloeckle, Sten Sootla, Itai Gat, Xiao- qing Ellen Tan, Yossi Adi, Jingyu Liu, Romain Sauvestre, Tal Remez, Jérémy Rapin, Artyom Kozhevnikov, Ivan Evtimov, Joanna Bitton, Manish Bhatt, Cris- tian Canton Ferrer, Aaron Grattafiori, Wenhan X...

2018 arXiv
[36]

Mukul Singh, José Cambronero, Sumit Gulwani, Vu Le, Carina Negreanu, Elnaz Nouri, Mohammad Raza, and Gust Verbruggen. 2023. FormaT5: Abstention and Examples for Conditional Table Formatting with Natural Language. Proc. VLDB Endow. 17, 3 (2023), 497–510. https://www.vldb.org/pv...

2023
[37]

Mukul Singh, José Cambronero, Sumit Gulwani, Vu Le, Carina Negreanu, and Gust Verbruggen. 2023. CodeFusion: A Pre-trained Diffusion Model for Code Generation. In Proceedings of the 2023 Conference on Empirical Meth- ods in Natural Language Processing , Houda Bouamor, Juan Pino...

2023 doi
[38]

Mukul Singh, José Cambronero, Sumit Gulwani, Vu Le, Carina Negreanu, and Gust Verbruggen. 2023. DataVinci: Learning Syntactic and Semantic String Repairs. arXiv:2308.10922 [cs.DB]

2023 arXiv
[39]

Mukul Singh, Rahul Kumar Dubey, and Swarup Kumar. 2022. Chapter 15 - Vehicle telematics: An Internet of Things and Big Data approach. In Artificial Intelligence and Machine Learning for EDGE Computing , Rajiv Pandey, Sunil Kumar Khatri, Neeraj kumar Singh, and Parul Verma (Eds...

2022
[40]

Mukul Singh, José Cambronero Sánchez, Sumit Gulwani, Vu Le, Carina Negreanu, Mohammad Raza, and Gust Verbruggen. 2023. Cornet: Learning Table Formatting Rules By Example. Proc. VLDB Endow. 16, 10 (jun 2023), 2632–2644. doi:10.14778 /3603581.3603600

2023
[41]

Mukul Singh, José Cambronero Sanchez, Sumit Gulwani, Vu Le, Carina Negreanu, and Gust Verbruggen. 2023. Cornet: Learning Spreadsheet Formatting Rules by Example. Proc. VLDB Endow. 16, 12 (aug 2023), 4058–4061. doi:10.14778/3611540 .3611620

2023 doi
[42]

Ananya Singha, Bhavya Chopra, Anirudh Khatry, Sumit Gulwani, Austin Henley, Vu Le, Chris Parnin, Mukul Singh, and Gust Verbruggen. 2024. Semantically Aligned Question and Code Generation for Automated Insight Generation. In Proceedings of the 1st International Workshop on Larg...

2024
[43]

Ananya Singha, Bhavya Chopra, Anirudh Khatry, Sumit Gulwani, Austin Henley, Vu Le, Chris Parnin, Mukul Singh, and Gust Verbruggen. 2024. Semantically Aligned Question and Code Generation for Automated Insight Generation. In LLM4Code Workshop at ICSE ’24 . https://www.microsoft...

2024
[44]

Kexuan Sun, Harsha Rayudu, and Jay Pujara. 2021. A Hybrid Probabilistic Approach for Table Understanding. Proceedings of the AAAI Conference on Artificial Intelligence 35, 5 (May 2021), 4366–4374. https://ojs.aaai.org/index.php /AAAI/article/view/16562

2021
[45]

Qiushi Sun, Nuo Chen, Jianing Wang, Xiang Li, and Ming Gao. 2023. TransCoder: Towards Unified Transferable Code Representation Learning Inspired by Human Skills. arXiv:2306.07285 [cs.SE]

2023 arXiv
[46]

W3C. 2020. CSS Color Module Level 3. https://www.w3.org/TR/css-color-3/

2020
[47]

Yue Wang, Hung Le, Akhilesh Deepak Gotmare, Nghi D. Q. Bui, Junnan Li, and Steven C. H. Hoi. 2023. CodeT5+: Open Code Large Language Models for Code Understanding and Generation. arXiv:2305.07922 [cs.CL]

2023 arXiv
[48]

Zhiruo Wang, Haoyu Dong, Ran Jia, Jia Li, Zhiyi Fu, Shi Han, and Dongmei Zhang. 2021. TUTA: Tree-Based Transformers for Generally Structured Table Pre-Training. In Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining (KDD ’21) . Association fo...

2021
[49]

Chi, Quoc V Le, and Denny Zhou

Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, brian ichter, Fei Xia, Ed H. Chi, Quoc V Le, and Denny Zhou. 2022. Chain of Thought Prompting Elicits Reasoning in Large Language Models. In Advances in Neural Information Pro- cessing Systems, Alice H. Oh, Alekh Agarwal,...

2022
[50]

Cong Yan and Yeye He. 2020. Auto-Suggest: Learning-to-Recommend Data Preparation Steps Using Data Science Notebooks. In International Conference on Management of Data (SIGMOD) . ACM, 1539–1554. https://www.microsoft.co m/en-us/research/publication/auto-suggest-learning-to-reco...

2020
[51]

Griffiths, Yuan Cao, and Karthik Narasimhan

Shunyu Yao, Dian Yu, Jeffrey Zhao, Izhak Shafran, Thomas L. Griffiths, Yuan Cao, and Karthik Narasimhan. 2023. Tree of Thoughts: Deliberate Problem Solving with Large Language Models. arXiv:2305.10601 [cs.CL]

2023 arXiv
[52]

Pengcheng Yin, Graham Neubig, Wen-tau Yih, and Sebastian Riedel. 2020. TaBERT: Pretraining for Joint Understanding of Textual and Tabular Data. In Proceedings of the 58th Annual Meeting of the Association for Computa- tional Linguistics. Association for Computational Linguisti...

2020 doi
[53]

Wei Zhao, Zhitao Hou, Siyuan Wu, Yan Gao, Haoyu Dong, Yao Wan, Hongyu Zhang, Yulei Sui, and Haidong Zhang. 2024. NL2Formula: Generating Spreadsheet Formulas from Natural Language Queries. arXiv:2402.14853 [cs.CL] https: //arxiv.org/abs/2402.14853

2024 arXiv
[2004]

Spreadsheets in Education 1 (2004), 85105

Spreadsheet Conditional Formatting: An Untapped Resource for Mathe- matics Education. Spreadsheets in Education 1 (2004), 85105

2004
[2020]

arXiv:1901.00409 [stat.ML]

Neural Clustering Processes. arXiv:1901.00409 [stat.ML]

1901 arXiv

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.

[1] [1]

Sergei Abramovich, Stephen Sugden, Sergei Abramovich, and Stephen J Sugden

work page

[2] [2]

Sibei Chen, Yeye He, Weiwei Cui, Ju Fan, Song Ge, Haidong Zhang, Dong- mei Zhang, and Surajit Chaudhuri. 2024. Auto-Formula: Recommend For- mulas in Spreadsheets using Contrastive Learning for Table Representations. arXiv:2404.12608 [cs.DB]

work page Pith review arXiv 2024

[3] [3]

Wenhu Chen, Xueguang Ma, Xinyi Wang, and William W. Cohen. 2023. Pro- gram of Thoughts Prompting: Disentangling Computation from Reasoning for Numerical Reasoning Tasks. Transactions on Machine Learning Research (2023). https://openreview.net/forum?id=YfZ4ZPt8zd

2023

[4] [4]

Xinyun Chen, Petros Maniatis, Rishabh Singh, Charles Sutton, Hanjun Dai, Max Lin, and Denny Zhou. 2021. SpreadsheetCoder: Formula Prediction from Semi- structured Context. In Proceedings of the 38th International Conference on Machine Learning (Proceedings of Machine Learning Research, Vol. 139) , Marina Meila and Tong Zhang (Eds.). PMLR, virtual, 1661–16...

work page 2021

[5] [5]

Haoyu Dong, Jinyu Wang, Zhouyu Fu, Shi Han, and Dongmei Zhang. 2020. Neural Formatting for Spreadsheet Tables. In Proceedings of the 29th ACM International Conference on Information & Knowledge Management (Virtual Event, Ireland) (CIKM ’20). Association for Computing Machinery, New York, NY, USA, 305–314. doi:10.1145/3340531.3411943

arXiv 2020

[6] [6]

Kevin Ellis and Sumit Gulwani. 2017. Learning to Learn Programs from Examples: Going Beyond Program Structure. In IJCAI 2017 (ijcai 2017 ed.). IJCAI 2017, Melbourne, Australia, 1638–1645. www.microsoft.com/research/publication/lea rning-learn-programs-examples-going-beyond-program-structure/

work page 2017

[7] [7]

Anna Fariha and Alexandra Meliou. 2019. Example-Driven Query Intent Discov- ery: Abductive Reasoning Using Semantic Similarity. Proc. VLDB Endow. 12, 11 (jul 2019), 1262–1275. doi:10.14778/3342263.3342266

arXiv 2019

[8] [8]

Anna Fariha, Ashish Tiwari, Alexandra Meliou, Arjun Radhakrishna, and Sumit Gulwani. 2021. CoCo: Interactive Exploration of Conformance Constraints for Data Understanding and Data Cleaning. In Proceedings of the 2021 Inter- national Conference on Management of Data (Virtual Event, China) (SIGMOD ’21). Association for Computing Machinery, New York, NY, USA...

arXiv 2021