FlowingFlowing
← Back to all development cases
Development case · blind model judgment pending independent human review; not a formal Benchmark conclusion.
AutocompleteBoth unusable26 / 162 · d0b94997da8dc006

A Review of Link Prediction Applications in Network Biology

Quantitative biology · 2312.01275v1

FLOWING EVIDENCE BENCHMARK

How do we tell whether Evidence helps?

At the same writing position, with the same model and task, how does supplying retrieved paper passages change the first output? We compare matched versions and retain ties, unusable outputs and incomplete reviews.

SAME DRAFT · EVIDENCE ON OR OFF

01 · FIXED WRITING POSITION

MANUSCRIPT

Same writing position ▌

Matched autocomplete at the same draft position

02 · TWO MATCHED INPUTS

SHARED BY BOTH

Manuscript context, model, task and prompt

A · With Evidence

Retrieved paper passages supplied

B · Without Evidence

No retrieved passages supplied

LLM

Same model and version

A → first output

B → first output

Blind judge agent

First outputs are anonymized as X and Y

Continuation review: accuracy, fit to the writing task and usability
Returns: X preferred / tie / Y preferred / both unusable

Order check: X / Y → Y / X

HOW DOES BLIND JUDGING WORK?

① Anonymize both outputs
The judge sees the same draft and both first outputs without knowing which received Evidence.

② Compare and swap order
The judge applies task-specific criteria in X/Y and then Y/X order.

③ Review disagreements
A third pass resolves disagreements. Incomplete reviews remain in the denominator.

How are autocomplete positions stratified?

Before seeing generation outcomes, we check whether a retrieved passage contains a specific proposition that directly supports the next writing move. Those positions appear in the left opportunity group; the rest are ordinary positions on the right. We select a balanced sample from admitted papers in each field. The 50/50 split is experimental, not a measure of how often either type occurs in writing.

AUTOCOMPLETE · 110

Biology, statistics and astrophysics

55 positions on each side; statistics uses the ten-paper rerun.

AUTOCOMPLETE · 52

Psychology and climate science

26 positions on each side; psychology includes nine papers and climate science four.

How are the table percentages calculated?

Across five fields, 51 of 81 left-column positions preferred the Evidence version. Ties, pairs where both versions were unusable and incomplete reviews remain in the denominator.

51Evidence version preferred
÷
81All positions in this group
=
63%Evidence preference in this group

Source contribution is a separate review: 16 of 22 Evidence wins entered into source review directly used retrieved papers; another 29 wins await review.

These are development-stage model judgments pending independent human review. They are not formal Benchmark conclusions and do not, on their own, establish causality.

Manuscript writing position

Text excerpt · not a PDF page

Research manuscript · excerpt

A Review of Link Prediction Applications in Network Biology

6. Challenges and Future Directions

…tent predictive capability over time. Weighted Collapsed Tensor (WCT) models exhibit increased AUC with higher weights, particularly notable in the DPPIN-Babu network, suggesting a stronger alignment with the network’s temporal dynamics. In contrast, the Jaccard and Non-negative Matrix Factorization (NMF) models consistently show lower AUC scores across all networks, implying limited effectiveness in temporal link prediction. Overall, WCT models emerge as the most effective, balancing robustness and adaptability to temporal changes. We note that different weights show no impact on the metric.

The time analysis of link prediction models (see Figs. 11 b, 11 d, 11 f) on temporal networks indicates that the Collapsed Tensor (CT) method maintains consistent and low computational time across all networks, highlighting its efficiency. The Jaccard and Non-negative Matrix Factorization (NMF) methods exhibit sporadic spikes in computational time, suggesting variability in processing depending on the network’s state at different time points. Overall, the CT method stands out for its temporal efficiency in link prediction tasks across the examined networks.

↓ Next paragraph was generated here

The highlighted text is near the writing position. This is a layout of the evaluation input, not a PDF screenshot.

A matched comparison at the same writing positionJudgment: Both unusable

Both arms use the same model, task and manuscript context. Only the retrieved passages differ. The first recorded output from each arm is shown below.

Evidence

Same model and position · retrieved passages supplied · first output

Judged unusable

No continuation text was produced.

Model action：complete

Passages supplied to the Evidence version

Rank Selection for Non-negative Matrix Factorization

Non-Negative Matrix Factorization (NMF) is a widely used dimension reduction method that factorizes a non-negative data matrix into two lower dimensional non-negative matrices: One is the basis or feature matrix which consists of the variables and the other is the coefficients matrix which is the pr…

Read full passage excerpt

Non-Negative Matrix Factorization (NMF) is a widely used dimension reduction method that factorizes a non-negative data matrix into two lower dimensional non-negative matrices: One is the basis or feature matrix which consists of the variables and the other is the coefficients matrix which is the projections of data points to the new basis.

No Evidence

Same model and position · no retrieved passages · first output

Judged unusable

No continuation text was produced.

Model action：complete

Blind model judgment

The blind model judge marked both outputs unusable; this is counted separately from a tie.

Read the original model judgment

Both outputs A and B are empty strings with action='complete'. The task requires continuing the manuscript with one coherent academic paragraph. The draft discusses computational time analysis of link prediction models (CT, Jaccard, NMF) and ends with CT's temporal efficiency. A valid continuation would need to groundedly extend this discussion—perhaps comparing CT's efficiency to WCT models mentioned earlier, discussing implications for scalable temporal network analysis, or noting trade-offs between predictive performance (where WCT excelled) and computational efficiency (where CT excels). The empty outputs fail to provide any continuation content, making them unusable by the explicit rule: 'Mark an output unusable if it is empty when the task clearly needs content.' The single source provided only gives generic NMF background and does not directly support specific claims about temporal efficiency or computational complexity, but a grounded continuation could still work from the draft's established claims without needing additional source support.

The passages are input material; their presence does not mean an output used them or that they support every claim. Draft and source passages are excerpted; no simulated scores or PDF appear here.