SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2410.04013 · NeurIPS · 2024

Improving Temporal Link Prediction via Temporal Walk Matrix Projection

Leilei Sun, Weifeng Lv, Xiaodong Lu, Tongyu Zhu

arXiv · PDF · Open in the Atlas

Code that ran

8 of 12 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 4 have no contract verdict; 4 failures of our instrument; 4 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/2410.04013.json · badge: /api/ran/2410.04013.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
lxd99/tpnet canonical (classifier label)
classifier basis: classifier_2025
4 ran · 3 no recorded run
lxd99/TPNet canonical (classifier label)
classifier basis: census
4 ran · 1 no recorded run
FunctionStatusWhere it lives
FeedForwardNet Ran
ran on a synthesized input
lxd99/tpnet/models/TPNet.py
code served (permissive licence) · get_code("b949112ded585c96")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned Tensor, ndim 2, dtype family float, shape (2, 4) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape 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 FeedForwardNet(nn.Module):

    def __init__(self, input_dim: int, dim_expansion_factor: float, dropout: float=0.0):
        """
        two-layered MLP with GELU activation function.
        :param input_dim: int, dimension of input
        :param dim_expansion_factor: float, dimension expansion factor
        :param dropout: float, dropout rate
        """
        super(FeedForwardNet, self).__init__()
        self.input_dim = input_dim
        self.dim_expansion_factor = dim_expansion_factor
        self.dropout = dropout
        self.ffn = nn.Sequential(nn.Linear(in_features=input_dim, out_features=int(dim_expansion_factor * input_dim)), nn.GELU(), nn.Dropout(dropout), nn.Linear(in_features=int(dim_expansion_factor * input_dim), out_features=input_dim), nn.Dropout(dropout))

    def forward(self, x: torch.Tensor):
        """
        feed forward net forward process
        :param x: Tensor, shape (*, input_dim)
        :return:
        """
        return self.ffn(x)


INIT_ARGS = {
    'input_dim': 4,
    'dim_expansion_factor': 2.5,
    'dropout': 0.1
}

ARGS = {
    'x': torch.randn(2, 4)
}

if __name__ == '__main__':
    model = FeedForwardNet(**INIT_ARGS)
    output = model(**ARGS)
    print(f"Input shape: {ARGS['x'].shape}")
    print(f"Output shape: {output.shape}")
    print(f"Output: {output}")
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 2, dtype family float, shape (2, 4) — shape (2, 4) as a later run recorded it
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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
MLPMixer Ran
ran on a synthesized input
lxd99/tpnet/models/TPNet.py
code served (permissive licence) · get_code("0783ca0f8dd2b03d")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned Tensor, ndim 3, dtype family float, shape (2, 5, 4) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape 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 FeedForwardNet(nn.Module):

    def __init__(self, input_dim: int, dim_expansion_factor: float, dropout: float=0.0):
        """
        two-layered MLP with GELU activation function.
        :param input_dim: int, dimension of input
        :param dim_expansion_factor: float, dimension expansion factor
        :param dropout: float, dropout rate
        """
        super(FeedForwardNet, self).__init__()
        self.input_dim = input_dim
        self.dim_expansion_factor = dim_expansion_factor
        self.dropout = dropout
        self.ffn = nn.Sequential(nn.Linear(in_features=input_dim, out_features=int(dim_expansion_factor * input_dim)), nn.GELU(), nn.Dropout(dropout), nn.Linear(in_features=int(dim_expansion_factor * input_dim), out_features=input_dim), nn.Dropout(dropout))

    def forward(self, x: torch.Tensor):
        """
        feed forward net forward process
        :param x: Tensor, shape (*, input_dim)
        :return:
        """
        return self.ffn(x)

class MLPMixer(nn.Module):

    def __init__(self, num_tokens: int, num_channels: int, token_dim_expansion_factor: float=0.5, channel_dim_expansion_factor: float=4.0, dropout: float=0.0):
        """
        MLP Mixer.
        :param num_tokens: int, number of tokens
        :param num_channels: int, number of channels
        :param token_dim_expansion_factor: float, dimension expansion factor for tokens
        :param channel_dim_expansion_factor: float, dimension expansion factor for channels
        :param dropout: float, dropout rate
        """
        super(MLPMixer, self).__init__()
        self.token_norm = nn.LayerNorm(num_tokens)
        self.token_feedforward = FeedForwardNet(input_dim=num_tokens, dim_expansion_factor=token_dim_expansion_factor, dropout=dropout)
        self.channel_norm = nn.LayerNorm(num_channels)
        self.channel_feedforward = FeedForwardNet(input_dim=num_channels, dim_expansion_factor=channel_dim_expansion_factor, dropout=dropout)

    def forward(self, input_tensor: torch.Tensor):
        """
        mlp mixer to compute over tokens and channels
        :param input_tensor: Tensor, shape (batch_size, num_tokens, num_channels)
        :return:
        """
        hidden_tensor = self.token_norm(input_tensor.permute(0, 2, 1))
        hidden_tensor = self.token_feedforward(hidden_tensor).permute(0, 2, 1)
        output_tensor = hidden_tensor + input_tensor
        hidden_tensor = self.channel_norm(output_tensor)
        hidden_tensor = self.channel_feedforward(hidden_tensor)
        output_tensor = hidden_tensor + output_tensor
        return output_tensor


INIT_ARGS = {
    'num_tokens': 5,
    'num_channels': 4,
    'token_dim_expansion_factor': 0.5,
    'channel_dim_expansion_factor': 4.0,
    'dropout': 0.1,
}

ARGS = {
    'input_tensor': torch.randn(2, 5, 4),
}

if __name__ == '__main__':
    model = MLPMixer(**INIT_ARGS)
    output = model(**ARGS)
    print(f"Output shape: {output.shape}")
    print(f"Expected shape: (2, 5, 4)")
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 3, dtype family float, shape (2, 5, 4) — shape (2, 5, 4) as a later run recorded it
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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
RandomProjectionModule Ran
ran on a synthesized input
lxd99/tpnet/models/TPNet.py
code served (permissive licence) · get_code("061020737965193c")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned NoneType. 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
import math

# INIT_ARGS for RandomProjectionModule.__init__
node_num = 10
edge_num = 20
dim_factor = 2
num_layer = 2
time_decay_weight = 0.1
device = 'cpu'
use_matrix = False
beginning_time = np.float64(0.0)
not_scale = False
enforce_dim = -1

INIT_ARGS = {
    'node_num': node_num,
    'edge_num': edge_num,
    'dim_factor': dim_factor,
    'num_layer': num_layer,
    'time_decay_weight': time_decay_weight,
    'device': device,
    'use_matrix': use_matrix,
    'beginning_time': beginning_time,
    'not_scale': not_scale,
    'enforce_dim': enforce_dim
}

# ARGS for the update method (which is the computation method)
batch_size = 3
src_node_ids = np.array([0, 2, 5], dtype=np.int64)
dst_node_ids = np.array([1, 3, 7], dtype=np.int64)
node_interact_times = np.array([1.0, 2.0, 3.0], dtype=np.float64)

ARGS = {
    'src_node_ids': src_node_ids,
    'dst_node_ids': dst_node_ids,
    'node_interact_times': node_interact_times
}

METHOD = 'update'
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type NoneType — a NoneType output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a NoneType 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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
TimeEncoder Ran
ran on a synthesized input
lxd99/tpnet/models/TPNet.py
code served (permissive licence) · get_code("0c789edc2dc5688c")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned Tensor, ndim 3, dtype family float, shape (2, 3, 8) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape 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
import torch.nn as nn

class TimeEncoder(nn.Module):

    def __init__(self, time_dim: int, parameter_requires_grad: bool=True):
        """
        Time encoder.
        :param time_dim: int, dimension of time encodings
        :param parameter_requires_grad: boolean, whether the parameter in TimeEncoder needs gradient
        """
        super(TimeEncoder, self).__init__()
        self.time_dim = time_dim
        self.w = nn.Linear(1, time_dim)
        self.w.weight = nn.Parameter(torch.from_numpy(1 / 10 ** np.linspace(0, 9, time_dim, dtype=np.float32)).reshape(time_dim, -1))
        self.w.bias = nn.Parameter(torch.zeros(time_dim))
        if not parameter_requires_grad:
            self.w.weight.requires_grad = False
            self.w.bias.requires_grad = False

    def forward(self, timestamps: torch.Tensor):
        """
        compute time encodings of time in timestamps
        :param timestamps: Tensor, shape (batch_size, seq_len)
        :return:
        """
        timestamps = timestamps.unsqueeze(dim=2)
        output = torch.cos(self.w(timestamps))
        return output


INIT_ARGS = {
    'time_dim': 8,
    'parameter_requires_grad': True
}

ARGS = {
    'timestamps': torch.randn(2, 3)
}

if __name__ == '__main__':
    encoder = TimeEncoder(**INIT_ARGS)
    result = encoder(**ARGS)
    print(f"Input shape: {ARGS['timestamps'].shape}")
    print(f"Output shape: {result.shape}")
    print(f"Output:\n{result}")
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 3, dtype family float, shape (2, 3, 8) — shape (2, 3, 8) as a later run recorded it
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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
compute_src_dst_node_time_shifts Ran
ran; our contract draft was wrong, not the code
lxd99/TPNet/models/MemoryModel.py
code served (permissive licence) · get_code("727d90514a9a5475")
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 tuple, 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 numpy as np

# Create small test data for the function
src_node_ids = np.array([0, 1, 0, 2, 1], dtype=np.int32)
dst_node_ids = np.array([1, 2, 3, 0, 1], dtype=np.int32)
node_interact_times = np.array([1.0, 2.5, 5.0, 3.2, 7.1], dtype=np.float32)

ARGS = {
    "src_node_ids": src_node_ids,
    "dst_node_ids": dst_node_ids,
    "node_interact_times": node_interact_times,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 1, dtype family float — a tuple 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
edge_bank_time_window_memory Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
lxd99/TPNet/models/EdgeBank.py
code served (permissive licence) · get_code("1de6e005e59cdf96")
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 set. 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 numpy as np

# Create small sample data for edge history
num_edges = 8

# Source and destination node IDs
history_src_node_ids = np.array([0, 1, 2, 3, 0, 1, 2, 3], dtype=np.int32)
history_dst_node_ids = np.array([1, 2, 3, 0, 2, 3, 0, 1], dtype=np.int32)

# Interaction times (increasing sequence)
history_node_interact_times = np.array([1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0], dtype=np.float32)

# Time window mode
time_window_mode = 'fixed_proportion'

# Time window proportion (keep last 50% of data)
time_window_proportion = 0.5

ARGS = {
    'history_src_node_ids': history_src_node_ids,
    'history_dst_node_ids': history_dst_node_ids,
    'history_node_interact_times': history_node_interact_times,
    'time_window_mode': time_window_mode,
    'time_window_proportion': time_window_proportion,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type set — a set output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a set 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
edge_bank_unlimited_memory Ran
ran; our contract draft was wrong, not the code
lxd99/TPNet/models/EdgeBank.py
code served (permissive licence) · get_code("0f56e3ae8a152b2b")
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 set. 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 numpy as np

# Create small test arrays for historical edge data
# history_src_node_ids: source node IDs for historical edges
# history_dst_node_ids: destination node IDs for historical edges

history_src_node_ids = np.array([0, 1, 2, 3, 0, 1], dtype=np.int64)
history_dst_node_ids = np.array([1, 2, 3, 0, 2, 3], dtype=np.int64)

ARGS = {
    "history_src_node_ids": history_src_node_ids,
    "history_dst_node_ids": history_dst_node_ids,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type set — a set output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a set 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
predict_link_probabilities Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
lxd99/TPNet/models/EdgeBank.py
code served (permissive licence) · get_code("8bcef3dc22f0489a")
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 ndarray, ndim 1, dtype family float, shape (5,) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape 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 numpy as np

# edge_memories: a set of tuples representing edges that exist in memory
# Each edge is a tuple of (src_node_id, dst_node_id)
edge_memories = {
    (0, 1),
    (1, 2),
    (2, 3),
    (0, 3),
}

# edges_tuple: a tuple of (src_node_ids, dst_node_ids)
# Both are sequences (lists or arrays) of node IDs
src_node_ids = [0, 1, 2, 3, 5]
dst_node_ids = [1, 2, 3, 0, 6]
edges_tuple = (src_node_ids, dst_node_ids)

ARGS = {
    "edge_memories": edge_memories,
    "edges_tuple": edges_tuple,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 1, dtype family float, shape (5,) — shape (5,) as a later run recorded it
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
NeighborSampler Not yet run
unverified (harvested, no recorded run)
lxd99/tpnet/models/TPNet.py
code served (permissive licence) · get_code("0c735a99a03bbb86")
How it ran

no invocation recorded.

TPNet Not yet run
unverified (harvested, no recorded run)
lxd99/tpnet/models/TPNet.py
code served (permissive licence) · get_code("a6c3b6089e63786c")
How it ran

no invocation recorded.

TPNetEmbedding Not yet run
unverified (harvested, no recorded run)
lxd99/tpnet/models/TPNet.py
code served (permissive licence) · get_code("c849574271a0c04e")
How it ran

no invocation recorded.

get_layer_type Not yet run
unverified (harvested, no recorded run)
lxd99/TPNet/models/NAT.py
code served (permissive licence) · get_code("7878b72c4ebda48a")
How it ran

no invocation recorded.

Also present in other papers' code

Of 12 samples lifted from this paper's linked repositories, 12 carry a function identity and 0 do not. Of the 12 distinct functions with an identity, 6 are also present in other papers whose code we hold (6 with at least one paper dated earlier, 0 the same month, 5 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
compute_src_dst_node_time_shifts 5 other papers
4 dated earlier, 1 later
2307.01026 (2023-07)
Temporal Graph Benchmark for Machine Learning on Temporal Graphs
2307.12510 (2023-07)
An Empirical Evaluation of Temporal Graph Benchmark
2405.17473 (2024-05)
Repeat-Aware Neighbor Sampling for Dynamic Graph Learning
and 2 more not named here
edge_bank_time_window_memory 5 other papers
4 dated earlier, 1 later
2307.01026 (2023-07)
Temporal Graph Benchmark for Machine Learning on Temporal Graphs
2307.12510 (2023-07)
An Empirical Evaluation of Temporal Graph Benchmark
2405.17473 (2024-05)
Repeat-Aware Neighbor Sampling for Dynamic Graph Learning
and 2 more not named here
edge_bank_unlimited_memory 5 other papers
4 dated earlier, 1 later
2307.01026 (2023-07)
Temporal Graph Benchmark for Machine Learning on Temporal Graphs
2307.12510 (2023-07)
An Empirical Evaluation of Temporal Graph Benchmark
2405.17473 (2024-05)
Repeat-Aware Neighbor Sampling for Dynamic Graph Learning
and 2 more not named here
predict_link_probabilities 5 other papers
4 dated earlier, 1 later
2307.01026 (2023-07)
Temporal Graph Benchmark for Machine Learning on Temporal Graphs
2307.12510 (2023-07)
An Empirical Evaluation of Temporal Graph Benchmark
2405.17473 (2024-05)
Repeat-Aware Neighbor Sampling for Dynamic Graph Learning
and 2 more not named here
TimeEncoder
brick of this paper's assembly: helper
2 other papers
1 dated earlier, 1 later · 1 only as an assembly brick
2303.13047 (2023-03)
Towards Better Dynamic Graph Learning: New Architecture and Unified Library
2605.24971 (2026-05)
TGFormer: Towards Temporal Graph Transformer with Auto-Correlation Mechanism
get_layer_type 1 other paper
1 dated earlier
2209.01084 (2022-09)
Neighborhood-aware Scalable Temporal Network Representation Learning

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

Temporal link prediction, aiming at predicting future interactions among entities based on historical interactions, is crucial for a series of real-world applications. Although previous methods have demonstrated the importance of relative encodings for effective temporal link prediction, computational efficiency remains a major concern in constructing these encodings. Moreover, existing relative encodings are usually constructed based on structural connectivity, where temporal information is seldom considered. To address the aforementioned issues, we first analyze existing relative encodings and unify them as a function of temporal walk matrices. This unification establishes a connection between relative encodings and temporal walk matrices, providing a more principled way for analyzing and designing relative encodings. Based on this analysis, we propose a new temporal graph neural network called TPNet, which introduces a temporal walk matrix that incorporates the time decay effect to simultaneously consider both temporal and structural information. Moreover, TPNet designs a random feature propagation mechanism with theoretical guarantees to implicitly maintain the temporal walk matrices, which improves the computation and storage efficiency. Experimental results on 13 benchmark datasets verify the effectiveness and efficiency of TPNet, where TPNet outperforms other baselines on most datasets and achieves a maximum speedup of 33.3× compared to the SOTA baseline. Our code can be found at https://github.com/lxd99/TPNet. Many real-world dynamic systems can be abstracted as a temporal graph [1], where entities and interactions among them are denoted as nodes and edges with timestamps respectively. Temporal link prediction, aiming at predicting future interactions based on historical interactions, is a fundamental task for temporal graph learning, which can not only help us understand the evolution pattern of the temporal graph but also is crucial for a series of real-world tasks such as recommendations for online platforms [2, 3] and information diffusion prediction [4,5]. Relative encodings have become an indispensable module for effective temporal link prediction [6][7][8][9] where, without them, node representations computed independently by neighbor aggregation will fail to capture the pairwise information. As the toy example shown in Figure 1, A and F will have the same node representation due to sharing the same local structure. Thus it can not be determined whether 38th Conference on Neural Information Processing Systems (NeurIPS 2024).

For agents

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

get_harvested_code_for_paper("2410.04013")
get_code_for_paper("2410.04013")
have("2410.04013")

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

curl https://syntology.ai/api/ran/2410.04013.json

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

[![Syntology run record](https://syntology.ai/api/ran/2410.04013.svg)](https://syntology.ai/paper/2410.04013)

Connect an agent — have() is free.