Bibliographic record
Abstract
Purpose Stemming is a critical preprocessing phase in information retrieval (IR) and natural language processing (NLP) tasks, aimed at reducing words to their root forms to improve query matching and retrieval performance. While effective stemming algorithms exist for high-resource languages, low-resource languages such as Gujarati lack a robust solution. Existing rule-based, dictionary-based and hybrid stemming techniques for the Gujarati language struggle to handle morphological variations, contextual understanding and out-of-vocabulary words, which limits their effectiveness in search engines and text-processing applications. Design/methodology/approach This research study proposes two distinct stemming techniques for Gujarati: (1) an IndoWordNet-based method and (2) a novel large language model (LLM)-based approach comprising three variants: word-level (LLM-WL), sentence-level (LLM-SL) and part of speech-sentence-level (LLM-POS-SL). For LLM implementation, we used the GPT-4-0613 model. These methods were evaluated on a curated Gujarati corpus of 20,849 words using standard metrics: precision, recall and F-score. The impact of stemming on IR performance was assessed using Gujarati Wikipedia search queries, with evaluation metrics including mean average precision (MAP) and Precision@10. Findings The LLM-POS-SL variant achieved the best results, with an average precision of 96.7%, a recall of 91.95% and an F-score of 94.26%. IR experiments further confirmed that LLM-POS-SL enhances search relevance and ranking with superior MAP and Precision@10. Practical implications These findings underscore the potential of LLM-based stemming in enhancing Gujarati NLP tools, improving the effectiveness of digital search systems. This research advances socio-technical development in regional-language IR and promotes more equitable access to information. Future work will focus on optimizing efficiency, scalability and domain-specific adaptations to further improve stemming accuracy. Originality/value To the best of our knowledge, this is the first reported study to leverage an LLM for the stemming task in Gujarati or any other Indian language.
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 | 0.001 | 0.005 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.004 | 0.003 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.002 | 0.002 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.003 | 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".