SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2503.18596 · EMNLP · 2025

LinkAlign: Scalable Schema Linking for Real-World Large-Scale Multi-Database Text-to-SQL

Yihan Wang, Xin Yang, Peiyu Liu

arXiv · PDF · Open in the Atlas

Code that ran

6 of 7 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 5 have no contract verdict; 1 failure of our instrument; 1 no recorded run), measured 2026-11-22; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2503.18596.json · badge: /api/ran/2503.18596.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
Satissss/LinkAlign canonical (classifier label)
classifier basis not recorded on the link
6 ran · 1 no recorded run
FunctionStatusWhere it lives
get_sql_files Ran
ran on a synthesized input
Satissss/LinkAlign/utils.py
code served (permissive licence) · get_code("f04fdf977dad5f7e")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned list, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import tempfile
import os
from pathlib import Path

# Create a temporary directory with some .sql files
temp_dir = tempfile.mkdtemp()

# Create a few test .sql files
Path(temp_dir, "query1.sql").touch()
Path(temp_dir, "query2.sql").touch()
Path(temp_dir, "script.sql").touch()

# Create some non-.sql files that should be ignored
Path(temp_dir, "readme.txt").touch()
Path(temp_dir, "data.csv").touch()

ARGS = {
    "directory": temp_dir,
    "suffix": ".sql"
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family float — a list output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
parse_json_from_str Ran
ran on a synthesized input
Satissss/LinkAlign/utils.py
code served (permissive licence) · get_code("a2a0e0eafb93e8ae")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned dict. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import json

def parse_json_from_str(string: str=None) -> dict:
    try:
        cleaned = string.translate(str.maketrans('', '', '\n`'))
        cleaned = cleaned.replace('json', '') if 'json' in cleaned else cleaned
        return json.loads(cleaned)
    except Exception as e:
        raise ValueError('Failed to parse JSON from string') from e

# Setup code to construct valid arguments
ARGS = {
    'string': '{"key": "value", "number": 42}'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
parse_list_from_str Ran
ran on a synthesized input
Satissss/LinkAlign/utils.py
code served (permissive licence) · get_code("f45a7407cacc0807")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned list, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

torch.manual_seed(0)

# The function takes a string parameter and parses it into a list
# It expects a string in format like "['a','b','c']" or similar variations
# The function cleans the string by removing quotes, brackets, spaces, newlines, backticks, and the word 'python'
# Then splits by comma

ARGS = {
    'string': "['item1','item2','item3']"
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family float — a list output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
process_data Ran
ran on a synthesized input
Satissss/LinkAlign/preprocess.py
code served (permissive licence) · get_code("55f2b77cb9cceeab")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned list, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import json
import os
import tempfile
from typing import Dict, List

# Create a temporary directory for the test
temp_dir = tempfile.mkdtemp()

# Build a minimal valid row dict that satisfies all the function's requirements
row = {
    'db_id': 'test_db_1',
    'table_names_original': ['users', 'orders'],
    'column_names_original': [
        (0, 'user_id'),
        (0, 'user_name'),
        (1, 'order_id'),
        (1, 'user_id')
    ],
    'column_descriptions': [
        (0, 'User identifier'),
        (0, 'User full name'),
        (1, 'Order identifier'),
        (1, 'Foreign key to users')
    ],
    'column_types': ['int', 'varchar', 'int', 'int'],
    'sample_rows': {
        'users': [
            {'user_id': 1, 'user_name': 'Alice'},
            {'user_id': 2, 'user_name': 'Bob'}
        ],
        'orders': [
            {'order_id': 101, 'user_id': 1},
            {'order_id': 102, 'user_id': 2}
        ]
    },
    'table_to_projDataset': {
        'users': {'projection': 'users_proj'},
        'orders': {'projection': 'orders_proj'}
    }
}

ARGS = {
    'row': row,
    'save_path': temp_dir,
    'exclude_db': None
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family float — a list output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
read_jsonl Ran
ran; our contract draft was wrong, not the code
Satissss/LinkAlign/preprocess.py
code served (permissive licence) · get_code("a22422413f1c610b")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned list, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import json
import tempfile
import os

# Create a temporary JSONL file for testing
temp_dir = tempfile.mkdtemp()
temp_file = os.path.join(temp_dir, "test.jsonl")

# Write sample JSONL data
with open(temp_file, 'w', encoding='utf-8') as f:
    f.write(json.dumps({"name": "Alice", "age": 30}) + "\n")
    f.write(json.dumps({"name": "Bob", "age": 25}) + "\n")
    f.write(json.dumps({"name": "Charlie", "age": 35}) + "\n")

ARGS = {
    "file_path": temp_file
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family float — a list output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
transform_name Ran
ran on a synthesized input
Satissss/LinkAlign/preprocess.py
code served (permissive licence) · get_code("91344f87d2ff63cf")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned str. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# The function transform_name takes two string parameters and performs string operations on them
# It concatenates them, truncates to 100 chars, and replaces certain special characters with underscores

ARGS = {
    'table_name': 'users',
    'col_name': 'email_address'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
load_db_size Not yet run
unverified (harvested, no recorded run)
Satissss/LinkAlign/GenerateSchemas.py
code served (permissive licence) · get_code("73bff840540fea62")
How it ran

no invocation recorded.

Also present in other papers' code

Of 7 samples lifted from this paper's linked repositories, 7 carry a function identity and 0 do not. Of the 7 distinct functions with an identity, 1 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 0 the same month, 0 later) and 6 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

FunctionAlso present inEarliest-datedOthers named
read_jsonl 1 other paper
1 dated earlier
2410.13509 (2024-10)
RAG-DDR: Optimizing Retrieval-Augmented Generation Using Differentiable Data Rewards

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Schema linking is a critical bottleneck in applying existing Text-to-SQL models to realworld, large-scale, multi-database environments. Through error analysis, we identify two major challenges in schema linking: (1) Database Retrieval: accurately selecting the target database from a large schema pool, while effectively filtering out irrelevant ones; and (2) Schema Item Grounding: precisely identifying the relevant tables and columns within complex and often redundant schemas for SQL generation. Based on these, we introduce LinkAlign, a novel framework tailored for large-scale databases with thousands of fields. LinkAlign comprises three key steps: multiround semantic enhanced retrieval and irrelevant information isolation for Challenge 1, and schema extraction enhancement for Challenge 2. Each stage supports both Agent and Pipeline execution modes, enabling balancing efficiency and performance via modular design. To enable more realistic evaluation, we construct AmbiDB, a synthetic dataset designed to reflect the ambiguity of real-world schema linking. Experiments on widely-used Text-to-SQL benchmarks demonstrate that LinkAlign consistently outperforms existing baselines on all schema linking metrics. Notably, it improves the overall Text-to-SQL pipeline and achieves a new state-of-the-art score of 33.09% on the Spider 2.0-Lite benchmark using only open-source LLMs, ranking first on the leaderboard at the time of submission. The codes are available at https://github.com/Satissss/LinkAlign.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2503.18596")
get_code_for_paper("2503.18596")
have("2503.18596")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2503.18596.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2503.18596.svg)](https://syntology.ai/paper/2503.18596)

Connect an agent — have() is free.