Evaluation of Short Answers Using Domain Specific Embedding and Siamese Stacked BiLSTM with Contrastive Loss
Bibliographic record
Abstract
Automatic short answer evaluation is the most complex task to perform as compared to evaluation of multiple choices and true or false type questions.Short descriptive answer tries to capture the overall knowledge gained by student related to the course, his remembering and presentation capabilities of the same.But sometimes the evaluation of such short descriptive answer becomes cumbersome and time consuming.So in this study, we are trying to address the issue of automated evaluation of short descriptive answers for Data Structures course by proposing a Siamese stacked Bidirectional LSTM neural network.The model utilizes the domain specific embedding generated by training gensim Word2Vec model on Data Structures domain.Domain specific embedding helps to identify the context relevant to domain which is difficult to understand in pre-trained embedding's due to ambiguity in words.The proposed model is trained using contrastive loss function and finally evaluation is made to determine whether student answer is correct or incorrect based on model answer provided by evaluator.The proposed architecture is tested using widely used Mohler's dataset and the results obtained are compared to baseline approaches using Pearson correlation coefficient and RMSE score.Also the proposed architecture is utilized on the dataset generated for specifically Data Structures course.For Mohler's dataset proposed framework achieves the best Pearson correlation value 0.668 compared to related baseline approaches.The results obtained has shown that proposed architecture is effective in investigating the relationship between complex descriptive sentences and performs the task of evaluation more similar to that of human evaluator.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".