SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2005.00574 · ACL · 2020

Clinical Reading Comprehension: A Thorough Analysis of the emrQA Dataset

Huan Sun, Xiang Yue, Bernal Jimenez Gutierrez

arXiv · PDF · Open in the Atlas

Code that ran

4 of 4 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; 4 failures of our instrument; 0 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; 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/2005.00574.json · badge: /api/ran/2005.00574.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
xiangyue9607/CliniRC canonical (classifier label)
classifier basis: classifier_2025
4 ran · 0 no recorded run
FunctionStatusWhere it lives
clean_datasets Ran
ran; our contract draft was wrong, not the code
xiangyue9607/CliniRC/src/preprocessing.py
code served (permissive licence) · get_code("b566aa6891512b47")
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) 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
import os

# Create a temporary JSON file with the expected structure
temp_data = {
    "data": [
        {
            "title": "medication",
            "paragraphs": [
                {
                    "context": ["This is a test context about medication and treatment"],
                    "qas": [
                        {
                            "question": ["What medication is recommended?"],
                            "answers": [
                                {
                                    "text": "aspirin",
                                    "evidence": "aspirin is recommended",
                                    "answer_entity_type": "drug"
                                }
                            ]
                        }
                    ]
                }
            ]
        },
        {
            "title": "relations",
            "paragraphs": [
                {
                    "context": ["The patient took medication for treatment"],
                    "qas": [
                        {
                            "question": ["What is the relation?"],
                            "answers": [
                                {
                                    "text": "took",
                                    "evidence": "took medication",
                                    "answer_entity_type": "relation"
                                }
                            ]
                        }
                    ]
                }
            ]
        },
        {
            "title": "other",
            "paragraphs": []
        }
    ]
}

# Create a temporary file
temp_file = tempfile.NamedTemporaryFile(mode='w', suffix='.json', delete=False)
json.dump(temp_data, temp_file)
temp_file.close()

ARGS = {
    "filename": temp_file.name
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
tqdm — 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
prune_sentence Ran
ran; our contract draft was wrong, not the code
xiangyue9607/CliniRC/src/preprocessing.py
code served (permissive licence) · get_code("c8d275909a5c68a7")
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) 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 re

def prune_sentence(sent):
    sent = ' '.join(sent.split())
    sent = sent.replace('_', ' ')
    sent = sent.replace('&lt;', '<')
    sent = sent.replace('&gt;', '>')
    sent = sent.replace('&amp;', '&')
    sent = sent.replace('&quot;', '"')
    sent = sent.replace('&nbsp;', ' ')
    sent = sent.replace('&apos;', "'")
    sent = sent.replace('_', '')
    sent = sent.replace('"', '')
    sent = re.sub('\\s([?.!\\\',)"](?:\\s|$))', '\\1', sent)
    sent = re.sub('[?.!,]+(?=[?.!,])', '', sent)
    sent = ' '.join(sent.split())
    return sent

ARGS = {
    'sent': 'Hello world &lt;tag&gt; &amp; test_string with "quotes" and punctuation!!!'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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 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
split_datasets_into_subsets Ran
ran; our contract draft was wrong, not the code
xiangyue9607/CliniRC/src/preprocessing.py
code served (permissive licence) · get_code("2be4f39928f86ff4")
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) 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

# Construct a minimal dataset_json that matches the expected structure
dataset_json = {
    'data': [
        {
            'paragraphs': [
                {
                    'context': 'This is a sample context with some text content here.',
                    'qas': [
                        {
                            'question': 'What is this?',
                            'id': 'q1',
                            'answers': [
                                {
                                    'text': 'sample context'
                                },
                                {
                                    'text': 'This is a sample context with some text content here and more words that exceed the limit'
                                }
                            ]
                        },
                        {
                            'question': 'Where is it?',
                            'id': 'q2',
                            'answers': [
                                {
                                    'text': 'here'
                                }
                            ]
                        }
                    ]
                },
                {
                    'context': 'Another paragraph with different content for testing purposes.',
                    'qas': [
                        {
                            'question': 'What else?',
                            'id': 'q3',
                            'answers': [
                                {
                                    'text': 'Another paragraph'
                                }
                            ]
                        }
                    ]
                }
            ]
        },
        {
            'paragraphs': [
                {
                    'context': 'Third subset context with more information.',
                    'qas': [
                        {
                            'question': 'More questions?',
                            'id': 'q4',
                            'answers': [
                                {
                                    'text': 'Third subset'
                                }
                            ]
                        }
                    ]
                }
            ]
        }
    ]
}

ARGS = {
    'dataset_json': dataset_json,
    'max_answer_length': 20
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy, torch, tqdm — 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
subsample_dataset_random Ran
ran; our contract draft was wrong, not the code
xiangyue9607/CliniRC/src/sample_dataset.py
code served (permissive licence) · get_code("146d8151db44fb77")
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) 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 random

# Construct a minimal data_json structure that matches what subsample_dataset_random expects
data_json = {
    'version': '1.1',
    'data': [
        {
            'title': 'Article 1',
            'paragraphs': [
                {
                    'context': 'This is the first context paragraph.',
                    'qas': [
                        {
                            'question': 'What is this?',
                            'id': 'q1',
                            'answers': [{'text': 'context', 'answer_start': 20}]
                        },
                        {
                            'question': 'Who wrote this?',
                            'id': 'q2',
                            'answers': [{'text': 'author', 'answer_start': 0}]
                        },
                        {
                            'question': 'When was this written?',
                            'id': 'q3',
                            'answers': [{'text': 'today', 'answer_start': 5}]
                        },
                        {
                            'question': 'Where is this from?',
                            'id': 'q4',
                            'answers': [{'text': 'here', 'answer_start': 10}]
                        }
                    ]
                },
                {
                    'context': 'This is the second context paragraph.',
                    'qas': [
                        {
                            'question': 'What else?',
                            'id': 'q5',
                            'answers': [{'text': 'more', 'answer_start': 15}]
                        },
                        {
                            'question': 'Why?',
                            'id': 'q6',
                            'answers': [{'text': 'reason', 'answer_start': 8}]
                        }
                    ]
                }
            ]
        },
        {
            'title': 'Article 2',
            'paragraphs': [
                {
                    'context': 'Another context here.',
                    'qas': [
                        {
                            'question': 'How?',
                            'id': 'q7',
                            'answers': [{'text': 'method', 'answer_start': 3}]
                        }
                    ]
                }
            ]
        }
    ]
}

ARGS = {
    'data_json': data_json,
    'sample_ratio': 0.5,
    'seed': 55
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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

Also present in other papers' code

Of 4 samples lifted from this paper's linked repositories, 4 carry a function identity and 0 do not. Of the 4 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 4 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

Machine reading comprehension has made great progress in recent years owing to largescale annotated datasets. In the clinical domain, however, creating such datasets is quite difficult due to the domain expertise required for annotation. Recently, Pampari et al. (2018) tackled this issue by using expert-annotated question templates and existing i2b2 annotations to create emrQA, the first large-scale dataset for question answering (QA) based on clinical notes. In this paper, we provide an indepth analysis of this dataset and the clinical reading comprehension (CliniRC) task. From our qualitative analysis, we find that (i) emrQA answers are often incomplete, and (ii) emrQA questions are often answerable without using domain knowledge. From our quantitative experiments, surprising results include that (iii) using a small sampled subset (5%-20%), we can obtain roughly equal performance compared to the model trained on the entire dataset, (iv) this performance is close to human expert's performance, and (v) BERT models do not beat the best performing base model. Following our analysis of the emrQA, we further explore two desired aspects of CliniRC systems: the ability to utilize clinical domain knowledge and to generalize to unseen questions and contexts. We argue that both should be considered when creating future datasets. 1

For agents

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

get_harvested_code_for_paper("2005.00574")
get_code_for_paper("2005.00574")
have("2005.00574")

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

curl https://syntology.ai/api/ran/2005.00574.json

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

[![Syntology run record](https://syntology.ai/api/ran/2005.00574.svg)](https://syntology.ai/paper/2005.00574)

Connect an agent — have() is free.