SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2210.08189 · NeurIPS · 2022

Parameter-free Dynamic Graph Embedding for Link Prediction

Dongsheng Li, Peng Zhang, Jiahao Liu, Hansu Gu, Tun Lu, Ning Gu

arXiv · PDF · Open in the Atlas

Code that ran

3 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 1 has no contract verdict; 2 failures of our instrument; 2 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/2210.08189.json · badge: /api/ran/2210.08189.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
fudancisl/freegem community
Papers with Code archive
3 ran · 2 no recorded run
FunctionStatusWhere it lives
Attention Ran
ran on a synthesized input
fudancisl/freegem/next-interaction-prediction/process.py
pointer only (licence: NONE) · get_code("3c9763a6b2b73048")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, fudancisl/freegem when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

build_data Ran
ran; our contract draft was wrong, not the code
fudancisl/freegem/next-interaction-prediction/process.py
pointer only (licence: NONE) · get_code("991daf5043970987")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, fudancisl/freegem when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

iSVD Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
fudancisl/freegem/next-interaction-prediction/process.py
pointer only (licence: NONE) · get_code("5ad5db31845ce7fa")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, fudancisl/freegem when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

new_stage Not yet run
unverified (harvested, no recorded run)
fudancisl/freegem/next-interaction-prediction/process.py
pointer only (licence: NONE) · get_code("062d1d8b6ba19006")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, fudancisl/freegem when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

run Not yet run
unverified (harvested, no recorded run)
fudancisl/freegem/next-interaction-prediction/process.py
pointer only (licence: NONE) · get_code("590019bc7569c3b1")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, fudancisl/freegem when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

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

Dynamic interaction graphs have been widely adopted to model the evolution of user-item interactions over time. There are two crucial factors when modelling user preferences for link prediction in dynamic interaction graphs: 1) collaborative relationship among users and 2) user personalized interaction patterns. Existing methods often implicitly consider these two factors together, which may lead to noisy user modelling when the two factors diverge. In addition, they usually require time-consuming parameter learning with back-propagation, which is prohibitive for real-time user preference modelling. To this end, this paper proposes FreeGEM, a parameter-free dynamic graph embedding method for link prediction. Firstly, to take advantage of the collaborative relationships, we propose an incremental graph embedding engine to obtain user/item embeddings, which is an Online-Monitor-Offline architecture consisting of an Online module to approximately embed users/items over time, a Monitor module to estimate the approximation error in real time and an Offline module to calibrate the user/item embeddings when the online approximation errors exceed a threshold. Meanwhile, we integrate attribute information into the model, which enables FreeGEM to better model users belonging to some under represented groups. Secondly, we design a personalized dynamic interaction pattern modeller, which combines dynamic time decay with attention mechanism to model user short-term interests. Experimental results on two link prediction tasks show that FreeGEM can outperform the state-of-the-art methods in accuracy while achieving over 36X improvement in efficiency. All code and datasets can be found in https://github.com/FudanCISL/FreeGEM. In this paper, we propose a Parameter-Free Dynamic Graph EMbedding (FreeGEM) method for link prediction. Here, parameter-free means that we do not incorporate any parameters which need to be learned via back-propagation. As shown in Figure 1, FreeGEM consists of two key components: incremental graph embedding engine and personalized dynamic interaction pattern modeller. The incremental graph embedding engine takes both historical data and online stream data as input and generates user/item embeddings as the output, which exploits the collaborative relationship among users. Its core innovation is the proposed Online-Monitor-Offline architecture, which can achieve online embedding updates by solving closed-form solutions in real time and keep the approximation errors caused by online singular value decomposition (SVD) [2] within any predefined threshold. In the Offline step, we propose a frequency-aware preference matrix reconstruction method to alleviate the oversmoothing problem and an attribute-integrated SVD to alleviate the cold-start issue. Surprisingly, the integration of attribute information enables FreeGEM to better model users belonging to under represented groups. In the Online step, we convert the offline truncated SVD into an online SVD to generate embeddings in real time. In the Monitor module, we estimate the online approximation error in real time by analyzing the relationship between approximation error and the update of online algorithm. The personalized dynamic interaction pattern modeller is also a parameter-free component, which combines the dynamic time decay with attention mechanism to model user short-term interests. It takes the user and item embeddings as input and outputs the prediction results. Specifically, it selectively forgets the early interactions through dynamic time decay and hence focuses on more recent interactions for prediction. Then, it leverages the attention mechanism to capture user personalized dynamic interaction patterns over the "decayed" item embedding sequences. Experimental results on two link prediction tasks (future item recommendation and next interaction prediction) show that FreeGEM can substantially outperform the state-of-the-art link prediction methods in accuracy while achieving over 36X improvement in computational efficiency. Besides, our empirical studies also confirm that FreeGEM can alleviate the cold-start issue and achieve high robustness on very sparse data. Link prediction methods on dynamic interaction graphs are mostly based on TPP, RNN and GNN, etc., which need time-consuming training process. To the best of our knowledge, the proposed FreeGEM is the only parameter-free method in this task, with high accuracy and efficiency. Know-Evolve [28] and HTNE [41] model interactions as multivariate point processes and Hawkes processes, respectively. Wang et al. [30] propose a co-evolutionary process to model the co-evolving nature of users and items. Shchur et al. [22] propose to directly model the conditional distribution of inter-event times. DSPP [3] incorporates topology and long-term dependencies into the intensity function. RNN-based methods RRN [32] models user and item interaction sequences with separate RNNs. Time-LSTM [40] proposes time gates to represent the time intervals. LatentCross [1] incorporates contextual data into embeddings. DeepCoevolve [6] and JODIE [14] generate node embeddings using two intertwined RNNs. JODIE [14] can also estimate user embeddings trajectories. DeePRed [13] employs non-recursive mutual RNNs to model interactions. GNN-based methods TDIG-MPNN [4] captures the global and local information on the graph. DGCF [16] uses three update mechanisms to update users and items. SDGNN [27] takes the

For agents

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

get_harvested_code_for_paper("2210.08189")
get_code_for_paper("2210.08189")
have("2210.08189")

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

curl https://syntology.ai/api/ran/2210.08189.json

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

[![Syntology run record](https://syntology.ai/api/ran/2210.08189.svg)](https://syntology.ai/paper/2210.08189)

Connect an agent — have() is free.