BIRD-QA: A BERT-based Information Retrieval Approach to Domain Specific Question Answering
Bibliographic record
Abstract
During recent years, Question Answering (QA) systems have been widely used in many industries to provide round the clock online services to consumers from all over the world. The importance of such services became more evident during the pandemic in online medical services, education, training, marketing, system support and administration. Most of the existing systems apply simple rule-based QA strategy. Human-defined rules are used to apply pattern matching for extracting information from a given data or knowledge base to generate responses to user queries. However, rule-based pattern matching techniques are not intelligent enough to understand the context of the question to always generate appropriate responses and are static. In this work, we explored different data preprocessing strategies and BERT-style pre-trained models to build an information retrieval (IR)-based Domain specific QA framework named BIRD-QA, and created a domain specific knowledge base using website data of a university department. We implemented multiple variations of extended BERT and ALBERT-base models and validated our framework on reading comprehension task using the Stanford Question Answering Dataset (SQuAD) 1.1 and 2.0 datasets. Our extended ALBERT-based model achieved 75.4% Exact Match (EM) score and 78.8% F1 score. We also present a small feasibility test of our framework for departmental QA using data from a university website.
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 machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.003 | 0.007 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.002 |
| Bibliometrics | 0.004 | 0.002 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.001 | 0.004 |
| Open science | 0.003 | 0.002 |
| Research integrity | 0.002 | 0.002 |
| Insufficient payload (model declined to judge) | 0.005 | 0.003 |
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 source (direct Gemma or distilled Codex), 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".