Scribd
Upload
5 views2 pages

Compare

This section compares the performance of the proposed machine learning models with existing studies, highlighting their effectiveness in classifying questions based on Bloom's Taxonomy. The results indicate that the ensemble model performed competitively against notable benchmarks, achieving robust accuracy with both 2000 and 4000 samples. Overall, the study emphasizes the advantages of ensemble learning in handling diverse question types and cognitive levels in assessments.

Uploaded by

shamimahmed82245
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
5 views2 pages

Compare

This section compares the performance of the proposed machine learning models with existing studies, highlighting their effectiveness in classifying questions based on Bloom's Taxonomy. The results indicate that the ensemble model performed competitively against notable benchmarks, achieving robust accuracy with both 2000 and 4000 samples. Overall, the study emphasizes the advantages of ensemble learning in handling diverse question types and cognitive levels in assessments.

Uploaded by

shamimahmed82245
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

5.

5 Comparison with Existing Studies and Our Proposed Models

Performance Benchmarking Across Datasets

In this section, we compare the performance of our machine learning models with those
reported in previous research. The goal is to determine how well our method works, particularly
when combined with a prior dataset and actual university exam questions.

Below is a comparative table that includes:

●​ Benchmarked results from recent and notable papers,​

●​ Performance from our proposed models using 2000 and 4000 samples,​

●​ Results from other referenced models (from the image) wherever accuracy is available.​

Table 5.1: Comparative Accuracy of Bloom’s Taxonomy Question Classification Models


Author(s) Dataset Model / Method Accurac
y

Mallikarjuna Chindukuri & S. NCERT Ensemble of pre-trained 94.10%


Sivanesan (proposed) models

Yahya et al. Ensemble of pre-trained 92.77%


models

Jain et al. Ensemble of pre-trained 88.50%


models

CLO Ensemble of pre-trained 78.44%


models

Yahya et al. (2012) - TF + SVM 92.30%

Das et al. (2020) - BERT (lr = 0.00003) 89.70%

Laddha et al. (2021) - CNN 80.00%

- LSTM 71.00%

Waheed et al. (2021) 1st dataset BloomNet 87.50%

2nd dataset BloomNet 84.00%

Sharma et al. (2022) - BERT + Dense NN 81.10%


Zhang et al. (2021) - Not specified 59.20%

This Study 2000 dataset Model Ensembling 89.75%

BERT 88.50%

RoBERTa 88.20%

DistilBERT 88.50%

TextCNN 89.75%

This Study 4000 dataset Model Ensembling 88.93%

BERT 87.57%

RoBERTa 87.50%

DistilBERT 86.30%

TextCNN 81.99%

Analysis and Summary

Given the realistic and diverse nature of our dataset, it is clear from the table that our model
ensembling method is quite competitive. While the NCERT dataset yielded the highest result
(94.10%) due to its scale and structure, our combined dataset of real and research-based
questions still performed robustly.

●​ On 2000 data, our ensemble model matched and in some cases outperformed results
from other BERT-based and CNN models.​

●​ Our ensemble approach demonstrated good scalability and generalization on 4000


data, albeit a modest decline in accuracy.​

●​ Overall, our contextual models and ensemble strategy performed better than other
research that used standalone methods (TF+SVM, CNN, LSTM).​

This comparison demonstrates our model's ability to handle a variety of question kinds and
cognitive levels in authentic assessment settings, in addition to highlighting the value of
ensemble learning.

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy