SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2406.13375 · 2024

ALiiCE: Evaluating Positional Fine-grained Citation Generation

arXiv · PDF · Open in the Atlas

Code that ran

5 of 5 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; 5 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/2406.13375.json · badge: /api/ran/2406.13375.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
ylXuu/ALiiCE canonical (classifier label)
classifier basis: classifier_2025
5 ran · 0 no recorded run
FunctionStatusWhere it lives
find_node_by_id Ran
ran; our contract draft was wrong, not the code
ylXuu/ALiiCE/myparser.py
code served (permissive licence) · get_code("1a403c743b0950ca")
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 DependencyTreeNode. 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

class DependencyTreeNode:
    """ Dependency tree node, representing a word in the sentence. """

    def __init__(self, id=None, span=None, citation=None):
        self.id = id
        self.span = span
        self.children = []
        self.edge_values = {}
        self.citation = citation

    def add_child(self, child, edge_value):
        self.children.append(child)
        self.edge_values[child] = edge_value

    def remove_child(self, child):
        if child in self.children:
            self.edge_values.pop(child)
            self.children.remove(child)

    def __repr__(self) -> str:
        if self.citation:
            return f'{self.span}:{self.id}:{self.citation}'
        else:
            return f'{self.span}:{self.id}'

def find_node_by_id(node: DependencyTreeNode, id: int):
    """ Search the target node according to id """
    if node.id == id:
        return node
    for child in node.children:
        child_node = find_node_by_id(child, id)
        if child_node and child_node.id == id:
            return child_node
    return None

# Build a small dependency tree structure
root_node = DependencyTreeNode(id=0, span="root", citation="root_cite")

child1 = DependencyTreeNode(id=1, span="child1", citation="cite1")
child2 = DependencyTreeNode(id=2, span="child2", citation="cite2")

grandchild1 = DependencyTreeNode(id=3, span="grandchild1", citation="cite3")

root_node.add_child(child1, edge_value="nsubj")
root_node.add_child(child2, edge_value="obj")
child1.add_child(grandchild1, edge_value="det")

ARGS = {
    "node": root_node,
    "id": 3
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type DependencyTreeNode — a DependencyTreeNode output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a DependencyTreeNode 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
generate_tree_text Ran
ran; our contract draft was wrong, not the code
ylXuu/ALiiCE/myparser.py
code served (permissive licence) · get_code("bba3de48577322e8")
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 torch
import torch.nn as nn

class DependencyTreeNode:
    """ Dependency tree node, representing a word in the sentence. """

    def __init__(self, id=None, span=None, citation=None):
        self.id = id
        self.span = span
        self.children = []
        self.edge_values = {}
        self.citation = citation

    def add_child(self, child, edge_value):
        self.children.append(child)
        self.edge_values[child] = edge_value

    def remove_child(self, child):
        if child in self.children:
            self.edge_values.pop(child)
            self.children.remove(child)

    def __repr__(self) -> str:
        if self.citation:
            return f'{self.span}:{self.id}:{self.citation}'
        else:
            return f'{self.span}:{self.id}'

def generate_tree_text(node: DependencyTreeNode):
    """ Convert the subtree of node into text """
    output = {}
    output[node.id] = node.span
    for child in node.children:
        child_output = generate_tree_text(child)
        output = {**output, **child_output}
    return output

# Build a simple dependency tree structure
root = DependencyTreeNode(id=0, span="The", citation=None)
child1 = DependencyTreeNode(id=1, span="quick", citation=None)
child2 = DependencyTreeNode(id=2, span="fox", citation=None)
grandchild = DependencyTreeNode(id=3, span="brown", citation="source1")

root.add_child(child1, edge_value="det")
root.add_child(child2, edge_value="nsubj")
child1.add_child(grandchild, edge_value="amod")

ARGS = {
    "node": root
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — 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
get_messages Ran
ran; our contract draft was wrong, not the code
ylXuu/ALiiCE/generate.py
code served (permissive licence) · get_code("d3454cb82cfad8b3")
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

torch.manual_seed(0)

ARGS = {
    'prompt': 'What is the capital of France?',
    'instruction': 'You are ChatGPT, a large language model trained by OpenAI.'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
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
get_prompt Ran
ran; our contract draft was wrong, not the code
ylXuu/ALiiCE/generate.py
code served (permissive licence) · get_code("a9aab662c8ac6543")
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 logging

torch_manual_seed = __import__('torch').manual_seed(0)

# Construct the item dict with required keys
item = {
    'question': 'What is machine learning?',
    'docs': [
        {
            'title': 'Introduction to ML',
            'text': 'Machine learning is a subset of artificial intelligence.'
        },
        {
            'title': 'Deep Learning Basics',
            'text': 'Deep learning uses neural networks with multiple layers.'
        },
        {
            'title': 'Supervised Learning',
            'text': 'Supervised learning requires labeled training data.'
        },
        {
            'title': 'Unsupervised Learning',
            'text': 'Unsupervised learning finds patterns in unlabeled data.'
        },
        {
            'title': 'Reinforcement Learning',
            'text': 'Reinforcement learning uses rewards and penalties.'
        },
        {
            'title': 'Neural Networks',
            'text': 'Neural networks are inspired by biological neurons.'
        }
    ]
}

# Construct the prompt_dict with all required keys
prompt_dict = {
    'instruction': 'Answer the following question based on the provided documents.',
    'demo_sep': '\n---\n',
    'demo_prompt': '{INST}\n\nQuestion: {Q}\n\nDocuments:\n{D}\n\nAnswer: {A}',
    'doc_prompt': '[{ID}] Title: {T}\nContent: {P}'
}

ARGS = {
    'item': item,
    'prompt_dict': prompt_dict,
    'psg_num': 5,
    'doc_key': 'text'
}
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 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
init_tree Ran
ran; our contract draft was wrong, not the code
ylXuu/ALiiCE/myparser.py
code served (permissive licence) · get_code("54b2723d4145767a")
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 DependencyTreeNode. 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 numpy as np

torch.manual_seed(0)

# Construct valid arguments for init_tree function
# The function builds a dependency tree from dictionaries

# child_dict: maps node id to list of child ids
child_dict = {
    0: [1, 2],      # root (id=0) has children 1 and 2
    1: [3],         # node 1 has child 3
    2: [],          # node 2 has no children
    3: []           # node 3 has no children
}

# span_dict: maps node id to span (text representation)
span_dict = {
    0: "root",
    1: "word1",
    2: "word2",
    3: "word3"
}

# cite_dict: maps node id to citation (optional)
cite_dict = {
    0: "cite0",
    1: "cite1",
    3: "cite3"
    # node 2 intentionally omitted to test optional citations
}

# edge_dict: maps child id to edge value
edge_dict = {
    1: "nsubj",
    2: "obj",
    3: "det"
}

# root_id: the id of the root node
root_id = 0

ARGS = {
    "child_dict": child_dict,
    "span_dict": span_dict,
    "cite_dict": cite_dict,
    "edge_dict": edge_dict,
    "root_id": root_id
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type DependencyTreeNode — a DependencyTreeNode output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a DependencyTreeNode 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 5 samples lifted from this paper's linked repositories, 5 carry a function identity and 0 do not. Of the 5 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 5 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

Large Language Model (LLM) can enhance its credibility and verifiability by generating text with citations. However, existing research on citation generation is predominantly limited to sentence-level statements, neglecting the significance of positional fine-grained citations that can appear anywhere within sentences. To facilitate further exploration of the positional fine-grained citation generation, we propose ALiiCE, the first automatic evaluation framework for this task. Our method employs a dependency tree based approach to parse the sentence-level claim into atomic claims. Then ALiiCE evaluates citation quality using three metrics, including positional fine-grained citation recall, precision, and coefficient of variation of citation positions. We evaluate the positional fine-grained citation generation performance of several LLMs on long-form QA datasets. Our experiments and analyses demonstrate the effectiveness and reasonableness of ALiiCE. We offer our insights into the current advancements and future directions for the positional fine-grained citation generation task.

For agents

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

get_harvested_code_for_paper("2406.13375")
get_code_for_paper("2406.13375")
have("2406.13375")

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

curl https://syntology.ai/api/ran/2406.13375.json

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

[![Syntology run record](https://syntology.ai/api/ran/2406.13375.svg)](https://syntology.ai/paper/2406.13375)

Connect an agent — have() is free.