SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2607.04735 · 2026

Identifiability of Relational Queries in Multi-View Pretraining

Daniel Hernández, Bahadur Ratan, Thapa

arXiv · PDF · Open in the Atlas

Code that ran

3 of 6 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 0 have no contract verdict; 3 failures of our instrument; 3 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/2607.04735.json · badge: /api/ran/2607.04735.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
danielhz/query-identifiability canonical (classifier label)
classifier basis: census
3 ran · 3 no recorded run
FunctionStatusWhere it lives
collect Ran
ran; our contract draft was wrong, not the code
danielhz/query-identifiability/analysis/export_realworld_ml.py
code served (permissive licence) · get_code("6781fbbd1805d3be")
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
from pathlib import Path

# Create a temporary directory for results
temp_dir = tempfile.mkdtemp()
results_dir = Path(temp_dir)

# Create sample JSON result files
sample_results_1 = {
    "results": [
        {
            "dataset": "cifar10",
            "query": "query_a",
            "model": "resnet18",
            "test_balanced_accuracy": 0.85,
            "certified": True
        },
        {
            "dataset": "cifar10",
            "query": "query_a",
            "model": "resnet18",
            "test_balanced_accuracy": 0.87,
            "certified": True
        },
        {
            "dataset": "imagenet",
            "query": "query_b",
            "model": "vgg16",
            "test_balanced_accuracy": 0.75,
            "certified": False
        }
    ]
}

sample_results_2 = {
    "results": [
        {
            "dataset": "cifar10",
            "query": "query_a",
            "model": "resnet18",
            "test_balanced_accuracy": 0.86,
            "certified": True
        },
        {
            "dataset": "imagenet",
            "query": "query_b",
            "model": "vgg16",
            "test_balanced_accuracy": 0.76,
            "certified": False
        }
    ]
}

# Write sample JSON files
with open(results_dir / "e4_realworld_seed0.json", "w") as f:
    json.dump(sample_results_1, f)

with open(results_dir / "e4_realworld_seed1.json", "w") as f:
    json.dump(sample_results_2, f)

# Define ARGS
ARGS = {
    "results_dir": results_dir,
    "datasets": ["cifar10", "imagenet"]
}
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
gnnog_from_benchmark Ran
ran; our contract draft was wrong, not the code
danielhz/query-identifiability/models/gnn_og.py
code served (permissive licence) · get_code("b300b3381c8e93d2")
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 GNNOG. 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
import torch.nn as nn

torch.manual_seed(0)

# Create a minimal stand-in benchmark class
class MockConfig:
    def __init__(self, n_attrs, domain_size):
        self.n_attrs = n_attrs
        self.domain_size = domain_size

class MockBenchmark:
    def __init__(self, n_attrs, domain_size, n_overlaps):
        self.config = MockConfig(n_attrs, domain_size)
        # Create _augmented_overlaps: list of attribute sets for each overlap
        # Make them small and distinct
        self._augmented_overlaps = [
            [0, 1],           # first overlap: attrs 0,1
            [1, 2],           # second overlap: attrs 1,2
            [2],              # third overlap: attr 2 only
        ][:n_overlaps]

# Parameters
n_attrs = 4
domain_size = 3
n_overlaps = 3

bench = MockBenchmark(n_attrs, domain_size, n_overlaps)

# Build adjacency list from overlaps
# adj[a] = list of neighboring attributes for attribute a
adj = [[] for _ in range(n_attrs)]
for overlap in bench._augmented_overlaps:
    for a in overlap:
        for b in overlap:
            if a != b and b not in adj[a]:
                adj[a].append(b)

ARGS = {
    'bench': bench,
    'adj': adj,
    'hidden_dim': 32,
    'n_layers': 2,
}
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 GNNOG — a GNNOG output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a GNNOG 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_json Ran
ran; our contract draft was wrong, not the code
danielhz/query-identifiability/analysis/export_pgfdata.py
code served (permissive licence) · get_code("29244bb954637e6f")
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
import tempfile
from pathlib import Path

# Create a temporary JSON file with test data
with tempfile.NamedTemporaryFile(mode='w', suffix='.json', delete=False) as f:
    test_data = {
        "key1": "value1",
        "key2": [1, 2, 3],
        "key3": {"nested": "object"}
    }
    json.dump(test_data, f)
    temp_path = f.name

# Define ARGS with the path parameter
ARGS = {
    "path": Path(temp_path)
}
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
annotate_n_tuples Not yet run
unverified (harvested, no recorded run)
danielhz/query-identifiability/analysis/export_pgfdata.py
code served (permissive licence) · get_code("cb8bb6fcf5dcacf6")
How it ran

no invocation recorded.

export_witnesses Not yet run
unverified (harvested, no recorded run)
danielhz/query-identifiability/analysis/export_realworld_tables.py
code served (permissive licence) · get_code("1d2112024f7f53f7")
How it ran

no invocation recorded.

write_tex Not yet run
unverified (harvested, no recorded run)
danielhz/query-identifiability/analysis/export_realworld_ml.py
code served (permissive licence) · get_code("d393420a79431285")
How it ran

no invocation recorded.

Also present in other papers' code

Of 6 samples lifted from this paper's linked repositories, 6 carry a function identity and 0 do not. Of the 6 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 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.

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

When data sources are integrated through a shared interface, a downstream query may or may not be determined by what the interface exposes: two globally consistent worlds can agree on every shared attribute yet disagree on the query answer. This ambiguity is structural-a property of the interface design, not the data volume-and cannot be resolved by collecting more records or training a larger model. We formalize query identifiability for data integration under interface laws-functional dependencies that hold uniformly across all legal worlds rather than within a single instance-and prove three results. (i) A polynomial-time certificate (CheckCert) decides identifiability via attribute closure, and is exact on instances that actually expose any residual ambiguity (which we formally call closure-separable). (ii) Non-identifiable queries face an irreducible 1 /2 minimax error floor for any estimator using only interface evidence, bounding multi-view pretraining systems from below. (iii) A minimum-augmentation algorithm (Greedy-MinAug) finds the smallest set of interface additions to certify a query, reducing the problem to Set Cover (logarithmic approximation). Experiments on synthetic benchmarks, real integration datasets spanning three domains (scholarly, product, and restaurant), and schemas up to 10 3 attributes confirm CheckCert is exact, both algorithms run in single-digit milliseconds, and ML classifiers exhibit the predicted error floor and abrupt capability gains that augmentation produces.

For agents

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

get_harvested_code_for_paper("2607.04735")
get_code_for_paper("2607.04735")
have("2607.04735")

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

curl https://syntology.ai/api/ran/2607.04735.json

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

[![Syntology run record](https://syntology.ai/api/ran/2607.04735.svg)](https://syntology.ai/paper/2607.04735)

Connect an agent — have() is free.