Claims

On the FCE corpus the feature set performs comparably to prior work (r = 0.64 on test), though below one coherence-feature study

Claim · All claims
Evidence · 1 study (2 entries) · associational r2 · q2
Currency · newest study 2016 · not yet checked against its sources · not reviewed by a person

Subclaims

Reading the evidence codes

q is the design tier, i the impact, n the sample, and the kind with its r the rigour judged within that kind: what each code means.

This claim: no standardized effect is recorded in observations/ for it yet, so no benchmark reading is possible.

q2 i? The FCE regression model reached r = 0.64 and MAE 3.6 on FCE-TEST, statistically indistinguishable from Yannakoudakis et al. (2011) but significantly below Yannakoudakis and Briscoe (2012). → Sowmya Vajjala 2016

Evidence

Sowmya Vajjala 2016

Sowmya Vajjala. (2016). Automated assessment of non-native learner essays: Investigating the role of linguistic features. International Journal of Artificial Intelligence in Education. https://arxiv.org/abs/1612.00729

q2 · i? · associational · r2

Regression on the FCE corpus using the same train-test split as prior research (1141 training texts from year 2000, 97 test texts from 2001). Test correlation was 0.64 with "a MAE of 3.6".

"On testing with FCE-T EST (containing 97 texts), the model had a correlation of 0.64 and a MAE of 3.6."

Sowmya Vajjala 2016 (2)

Sowmya Vajjala. (2016). Automated assessment of non-native learner essays: Investigating the role of linguistic features. International Journal of Artificial Intelligence in Education. https://arxiv.org/abs/1612.00729

q2 · i? · associational · r2

Cocor correlation comparison against prior FCE results: no significant difference versus Yannakoudakis et al. (2011), but the Yannakoudakis and Briscoe (2012) result was "significantly better (p < 0.05)" than this model.

"However, the result from the second study (Yannakoudakis and Briscoe, 2012) is significantly better (p < 0.05) than the current model, which can perhaps be attribute d to the presence of additional coherence features in their model"

Discussion

Page metadata
Field Value
type claim
id fce-regression-comparable-to-prior-aes-work
title On the FCE corpus the feature set performs comparably to prior work (r = 0.64 on test), though below one coherence-feature study
description On the FCE corpus the feature set performs comparably to prior work (r = 0.64 on test), though below one coherence-feature study
status review
evidence_strength mixed
newest_evidence 2016
generated by process:wiki-ingest on 2026-10-27
trust unverified — no one has confirmed this page's content
sources 2 — sowmya-vajjala-2016, sowmya-vajjala-2016-2

Frontmatter is the page's machine-readable record — see the Schema & Guide.