LTEM: LLM-Friendly Table-Enhanced Markdown for Table Understanding and Generation

Eunbin Kim; Hyeonseok Kang; Sangkeun Jung

cs.CLOct 21, 2026

LTEM: LLM-Friendly Table-Enhanced Markdown for Table Understanding and Generation

Authors: Eunbin Kim, Hyeonseok Kang, Sangkeun Jung

Organizations: Department of Computer Science and Engineering, Chungnam National University, Republic of Korea · EurekaAI

Abstract

Tables in real-world documents, including scientific papers, financial reports, and statistical summaries, frequently use merged cells, multi-level headers, and sectioned layouts. Yet the lightweight formats LLMs handle best (Markdown, CSV) cannot express these structures, while expressive formats (HTML, LaTeX) impose excessive token cost and parsing complexity. To enable systematic analysis of this trade-off, we introduce a taxonomy decomposing table structure into six atomic features (F1-F6) and corresponding complexity levels (L1-L6), and construct STB (Synthetic Table Benchmark), a controlled benchmark with 1,089 tables systematically distributed across complexity levels. Using this framework, we propose LTEM (LLM-friendly Table-Enhanced Markdown), a formally specified, minimal extension of Markdown that explicitly encodes spans and hierarchical headers. Experiments across five state-of-the-art LLMs show that LTEM achieves accuracy comparable to HTML, offers a favorable accuracy-token trade-off relative to HTML and LaTeX, and produces substantially fewer invalid responses than LaTeX; for table generation, LTEM is comparable to MultiMarkdown though HTML is best. We release our taxonomy, benchmark, and format specification to support future research on LLM-table interaction.

Figures & tables

Feature	CSV	MD	MMD	HTML	LaTeX	LTEM	LTEM Syntax
F1 Grid Formation	✓	✓	✓	✓	✓	✓	\| \|
F2 Header Designation	$\triangle$	✓	✓	✓	✓	✓	\|---\|
F3 Cell Merging	✗	✗	$\triangle$	✓	✓	✓	<< , ^^ , {c:N,r:M}
F4 Header Nesting	✗	✗	$\triangle$	✓	✓	✓	multi-header rows
F5 Section Stacking	✗	✗	✗	$\triangle$	✓	✓	section-label rows
F6 Table Composition	✗	✗	✗	$\triangle$	✓	✓	\| # id \| , @ref

Table 1: Structural feature support by format. ✓ = native support;

\triangle

= partial or non-standard; ✗ = not supported. Dashed vertical rule separates existing formats from LTEM.

Panel of Figure 1 — Figure 1: Colspan in LTEM. (a) illustrates why explicit span markers matter; (b) shows marker- and attribute-based forms are structurally equivalent.

Figure 2: Multi-level headers in LTEM: multiple header rows above the separator line naturally encode hierarchical header nesting (F4).

Figure 3: Rich cell content in LTEM: inline line breaks ( \ ) enable structured content (lists, checkboxes) and math within a single cell.

Figure 4: Relational tables in LTEM: tables are decomposed and linked via explicit table IDs ( | # id | ) and references ( @ref ) instead of nesting.

Figure 5: Model $\times$ serialization-format performance (nEM) on STB. Format sensitivity varies by model; notably, Qwen3-235B exhibits a large degradation under LaTeX serialization.

Format	nEM $\uparrow$	TokenF1 $\uparrow$	NumEM $\uparrow$	PromptTok $\downarrow$	TableTok $\downarrow$	TNP ${}_{\text{adj}}\uparrow$
LTEM	0.813 ±0.002	0.844 ±0.002	0.943 ±0.000	449	228	1.809
HTML	0.812 ±0.001	0.844 ±0.000	0.945 ±0.000	612	392	1.326
MMD	0.810 ±0.000	0.843 ±0.000	0.946 ±0.001	436	215	1.858
MD	0.810 ±0.000	0.842 ±0.000	0.945 ±0.001	437	217	1.852
CSV	0.808 ±0.001	0.840 ±0.001	0.940 ±0.000	370	149	2.183
LaTeX	0.777 ±0.001	0.812 ±0.001	0.933 ±0.001	447	227	1.737

Table 2: Overall STB performance by format (average over 5 models, 3 runs). PromptTok excludes LTEM guide tokens; TNP

{}_{\text{adj}}

= nEM/PromptTok

\times

1000. LTEM and HTML achieve comparable nEM. Shorter formats obtain higher token-normalized scores, while LTEM provides a favorable accuracy–token trade-off relative to HTML and LaTeX.

Format	Refusal (%)	Invalid (%)	Total Error (%)
LTEM	0.15	0.14	0.29
CSV	0.22	0.09	0.31
HTML	0.30	0.11	0.41
MMD	0.34	0.15	0.49
MD	0.46	0.15	0.61
LaTeX	0.38	2.67	3.04

Table 3: Error rate analysis on STB. LTEM shows the lowest refusal rate. LaTeX exhibits a markedly higher invalid response rate than the other formats, primarily due to formatting and parsing failures.

Figure 6: Average performance (nEM) by structural complexity level (L1–L6) on STB. Arrow labels report absolute nEM changes in percentage points (pp). The highlighted L2–L3 and L3–L4 transitions correspond to the introduction of cell merging and multi-level headers.

Figure 7: Format-wise performance comparison (nEM) between STB and HiTab. The relative ordering of formats changes across datasets.

Format	ANLS $\uparrow$	Good+ (%)	PromptTok $\downarrow$
HTML	0.913	96.2	1149
LaTeX	0.749	72.2	1168
MultiMarkdown	0.734	69.0	1170
LTEM	0.720	67.9	1429
CSV	0.618	50.0	1170
Markdown	0.549	41.2	1168

Table 4: Table Generation performance (HiTab average; ANLS: normalized string similarity). PromptTok includes format-specific generation instructions, with an additional syntax guide for LTEM.

Appendix figures & tables17 assets

Supplementary material from the paper’s appendix.

Appendix

Level	Required Features	Description	Example
L1	F1	Atomic Grid	Simple data listing
L2	F1 + F2	Headered Table	Standard table with headers
L3	F1 + F2 + F3	Spanning Table	Table requiring cell merging
L4	F1 + F2 + F3 + F4	Multi-level Header	Hierarchical header structure
L5	F1 + F2 + F3 + F4 + F5	Block-partitioned	Same-schema section division
L6	F1 + F2 + F3 + F4 + F5 + F6	Composite Table	Multi-schema combination

Appendix

Table 5: Complexity Level Definitions.

Level	#Tables	Share (%)	Avg. rows	Avg. cols
L1 (Grid Only)	96	8.82	4.33	3.15
L2 (Basic Header)	334	30.67	5.63	3.34
L3 (Spanned)	342	31.40	7.15	5.22
L4 (Multi-level Header)	164	15.06	8.84	6.65
L5 (Section Stacking)	87	7.99	12.66	4.20
L6 (Composite)	66	6.06	15.55	11.50

Appendix

Table 6: Complexity-level distribution and average table size of STB (tables containing all four intents: LOOKUP/COUNTING/COMPARISON/AGGREGATION). Average rows/cols aggregate all sub-tables for multi-table cases.

Domain	#Tables	Share (%)
Scientific	366	33.61
Financial	286	26.26
Survey	179	16.44
Product	171	15.70
Schedule	87	7.99
Total	1,089	100.00

Appendix

Table 7: Domain distribution of STB (tables containing all four intents).

		FY2025		FY2026
Area	Program	Budget	Spent	Budget	Spent	Change
	E-document standards	1,240	1,180	1,380	612	+11.3%
	Open public data	860	845	910	402	+5.8%
Digital	Data structuring	-	-	420	95	-
	Regional hubs	2,100	2,064	2,050	1,010	-2.4%
Regional	Youth settlement	730	701	790	355	+8.2%

Appendix

Figure 9: A report table that combines row spans, column spans, and a two-row header (F3–F4), as LTEM source (top) and the rendered table (bottom). Numeric columns use the Markdown right-alignment marker ( ---: ).

Symbol	Meaning
=	Definition
,	Concatenation
\|	Alternation
[ ]	Optional (0 or 1 occurrence)
{ }	Repetition (0 or more occurrences)
( )	Grouping

Appendix

Table 8: Notation conventions used in the EBNF specification.

Property	Value	Implication
Chomsky Type	Type 2 (CFG)	Efficient parsing (O(n 3 ) worst, O(n) practical)
Ambiguity	Unambiguous	Deterministic parsing
Left Recursion	None	LL parser compatible
Lookahead	LL(1) for most constructs	Simple recursive descent
Unicode	Supported in content	UTF-8 encoding

Appendix

Table 9: Summary of grammar properties.

Format	nEM $\uparrow$	TokenF1 $\uparrow$	NumEM $\uparrow$	PromptTok $\downarrow$	TNP ${}_{\text{adj}}\uparrow$	Rank
HTML	0.295 ±0.001	0.331 ±0.002	0.519 ±0.001	1310	0.225	1
LaTeX	0.292 ±0.003	0.333 ±0.003	0.527 ±0.002	966	0.303	2
LTEM	0.289 ±0.005	0.329 ±0.004	0.504 ±0.004	1101	0.262	3
MD	0.285 ±0.003	0.322 ±0.004	0.515 ±0.003	1078	0.264	4
MMD	0.281 ±0.001	0.320 ±0.001	0.520 ±0.002	1058	0.266	5
CSV	0.266 ±0.004	0.306 ±0.003	0.515 ±0.003	770	0.345	6

Appendix

Table 10: Overall performance on the HiTab test split (515 hierarchical tables). HTML achieves the highest nEM, followed by LaTeX. The ranking of LaTeX reverses compared to STB (rank 6

\rightarrow

2), suggesting format effectiveness is dataset-dependent.

Format	Excellent (%)
HTML	76.2
LaTeX	29.3
MultiMarkdown	35.4
LTEM	25.8
CSV	25.4
Markdown	22.5

Appendix

Table 11: Excellent-rate breakdown for Table Generation on HiTab. Excellent denotes ANLS

\geq 0.9

; the values use the same evaluation setting as Table 4 .

Level	N	MD nEM	LTEM nEM	$\Delta$	Invalid MD/LTEM
L1	96	0.863	0.866	+0.003	0.05/0.05
L2	334	0.845	0.847	+0.002	0.22/0.13
L3	342	0.787	0.791	+0.004	0.11/0.16
L4	164	0.735	0.737	+0.002	0.17/0.19
L5	87	0.858	0.863	+0.004	0.00/0.00
L6	66	0.788	0.793	+0.005	0.28/0.20

Appendix

Table 12: Level-wise comparison of Markdown and LTEM on STB. Invalid rates are percentages.

Feature slice	N	MD nEM	LTEM nEM	$\Delta$	Invalid MD/LTEM
F2	1,057	0.809	0.812	+0.003	0.16/0.14
F3	593	0.783	0.787	+0.004	0.11/0.15
F4	345	0.762	0.766	+0.004	0.17/0.20
F5	87	0.858	0.863	+0.004	0.00/0.00
F6	66	0.788	0.793	+0.005	0.28/0.20

Appendix

Table 13: Feature-conditioned Markdown–LTEM comparison on STB. F2–F6 denote operational slices in which the corresponding structural feature is observed; the slices may overlap and are not cumulative level subsets. Invalid rates are percentages.

Element ( $N$ )	LTEM	Alternative	nEM (%)	p
Colspan (400)	<<	{c:N}	77.8 / 78.8	.481
Rowspan (348)	^^	{r:N}	75.6 / 76.7	.454
Header scope (288)	Separator	Header count	75.3 / 77.1	.302
Section label (348)	Marker row	Attribute row	85.6 / 85.9	1.000
Table ID (264)	In-grid row	MD heading	83.3 / 82.2	.453

Appendix

Table 14: Element-wise LTEM design-choice ablation on STB using GPT-5. Each row compares the LTEM form with a structure-preserving alternative on the same QA pairs (

N

). nEM is reported as LTEM/alternative; p -values are from exact McNemar tests.

Condition	Bucket	HTML	LaTeX	LTEM	MD
Merged density	None (430)	0.852	0.811	0.852	0.849
Merged density	Low (318)	0.781	0.750	0.781	0.783
Merged density	Medium (265)	0.792	0.759	0.796	0.786
Merged density	High (10)	0.735	0.725	0.725	0.725
Duplicate-value	None (668)	0.823	0.789	0.822	0.820
Duplicate-value	Low (246)	0.773	0.741	0.778	0.775

Appendix

Table 15: Structure-conditioned nEM on STB L1–L5 by merged-cell density and duplicate-value ambiguity. Table counts are shown in parentheses.

Format	nEM	Refusal	Invalid	Logged error
CSV	0.808	0.22	0.09	0.11
HTML	0.812	0.30	0.11	0.15
LaTeX	0.777	0.38	2.67	3.55
LTEM	0.813	0.15	0.14	0.18
MD	0.810	0.46	0.15	0.20
MMD	0.810	0.34	0.15	0.19

Appendix

Table 16: Failure diagnostics on STB. Refusal and Invalid are calculated among nEM-incorrect outputs; Logged error is recorded independently and may overlap. Values other than nEM are percentages.

Format	Aggregation	Comparison	Counting	Lookup
CSV	0.529	0.949	0.919	0.834
HTML	0.531	0.954	0.925	0.838
LaTeX	0.513	0.915	0.891	0.788
LTEM	0.530	0.954	0.925	0.843
MD	0.529	0.955	0.923	0.831
MMD	0.532	0.955	0.924	0.830

Appendix

Table 17: Question-type nEM breakdown on STB.

Backbone	Base nEM	LoRA nEM	$\Delta$ nEM
Qwen3-4B-Instruct	0.489	0.763	+0.274
Phi-4-mini	0.539	0.737	+0.198
Gemma-3-4B-IT	0.361	0.707	+0.346

Appendix

Table 18: LTEM LoRA pilot results on held-out STB tables using normalized Exact Match (nEM) as defined in the main evaluation.

Slice	Base nEM	LoRA nEM	$\Delta$
LOOKUP	0.817	0.983	+0.165
COUNTING	0.609	0.696	+0.087
AGGREGATION	0.148	0.478	+0.330
COMPARISON	0.383	0.896	+0.513
L1	0.500	0.773	+0.273
L2	0.522	0.750	+0.228

Appendix

Table 19: Qwen3-4B LTEM LoRA breakdown by question type and complexity level.

Model	n 01	n 10	Fisher p	Winner
DeepSeek-R1	196	343	5.4 $\times$ 10 -9 ***	LTEM
Llama-4-Maverick	343	441	0.0056**	LTEM
Gemini-2.5-Pro	253	275	0.410 (ns)	—
GPT-5	246	353	2.1 $\times$ 10 -4 ***	LTEM
Qwen3-235B	398	2,377	$<$ 10 -100 ***	LTEM

Appendix

Table 20: McNemar test comparing LTEM vs. LaTeX on STB. n 01 denotes LTEM-wrong & LaTeX-correct, and n 10 denotes LTEM-correct & LaTeX-wrong. The discordant counts are summed across three runs, and the p-values are Fisher-combined. Significance: *** p

<

0.001, ** p

<

0.01, * p

<

0.05.

Document Understanding

Document understanding enables machines to interpret text, layout, tables, and images to extract information and answer questions about documents.

See all 93 papers

TABVERSE: Benchmarking Cross-Format Table Understanding in LLMs and VLMs
Jun 8, 2026Momina Ahsan, Sarfraz Ahmad, Ming Shan Hee +2Document Understanding Table QA
1Mohamed bin Zayed University of Artificial Intelligence (MBZUAI) · 2Singapore University of Technology and Design (SUTD)
HighlightBench: Benchmarking Markup-Driven Table Reasoning in Scientific Documents
Mar 25, 2026Lexin Wang, Shenghua Liu, Yiwei Wang +5Multimodal Model Evaluation Table QA
Institute of Computing Technology, Chinese Academy of Sciences · University of California, Merced · University of Queensland
TILE: Bridging LLMs and Table Foundation Models for Tabular Learning with Strings
Oct 12, 2026Félix Lefebvre, David Holzmüller, Gioia Blayer +3Tabular ML Tabular Representation Learning
SODA Team, Inria Saclay · SAP SE · Probabl

Jun 8, 2026J/K move · Enter open · S save

TABVERSE: Benchmarking Cross-Format Table Understanding in LLMs and VLMs

Momina Ahsan, Sarfraz Ahmad, Ming Shan Hee, Roy Ka-Wei Lee+1

1Mohamed bin Zayed University of Artificial Intelligence (MBZUAI) · 2Singapore University of Technology and Design (SUTD)

Document Understanding Table QA LLM Evaluation Table Structure Recognition Multimodal Model Evaluation Multimodal Understanding

Open paper AskPDF arXiv

Large Language Models (LLMs) and Vision-Language Models (VLMs) are increasingly evaluated on table reasoning tasks, but the role of table representation remains under-explored. In practice, the same table content may appear in different structural formats, such as HTML, Markdown, and LaTeX, or as rendered images. However, existing evaluations often let content, format, layout, and modality vary together, making it difficult to isolate representation effects. We introduce TABVERSE, a controlled multimodal table benchmark that aligns the same table content across multiple structural formats and rendered images, with question category and difficulty tags. This design enables systematic evaluation of representation effects while holding table content fixed. We evaluate LLMs and VLMs across three tasks: Question Answering (QA), Structural Understanding Capability (SUC), and Structure Reconstruction (SR). Our results show that representation choice substantially affects table understanding. Models generally perform better with structured text than with rendered images, but the size of this gap depends on the task, model, and format. HTML is often the most robust text format, while row-sensitive structural tasks and syntactically usable LaTeX reconstruction remain challenging. These findings show that table representation is a key factor in reliable table evaluation.

Abstract

Document Understanding

Explore similar work

TABVERSE: Benchmarking Cross-Format Table Understanding in LLMs and VLMs

HighlightBench: Benchmarking Markup-Driven Table Reasoning in Scientific Documents

TILE: Bridging LLMs and Table Foundation Models for Tabular Learning with Strings