First-author NLP researcher: explainable AI (SHAP/LIME) for right-to-left scripts and hate speech detection.
2
papers here
2
reviewed by Evolve
2
first-author
0
corresponding
—
works · OpenAlex
—
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-01) and cover the author's entire public record, not just the papers listed here. Reproducibility signal comes from Evolve's own reviews.
Research profile
Output—
Not yet scored: this axis has no underlying data in the current snapshot. It would measure the volume of research outputs and their venues; so far only two arXiv preprints from 2026 are cataloged, which is too few for an output score.
Impact—
No citation or uptake data is available yet, so impact is unscored. This axis would measure citations, downstream use, and community influence of the work; both cataloged items are recent 2026 arXiv preprints with no citation records in the snapshot.
Focus0/100
The focus dimension scores 0, and the raw field distribution explains it: the two cataloged papers split exactly 0.5/0.5 between explainable-AI/NLP visualization and natural language processing, so no field dominates in a two-paper record. Both works are adjacent XAI-flavored NLP, however, so a clearer focus could consolidate as the catalog grows.
Collaboration28/100
Collaboration scores 28, consistent with a compact footprint: 3 distinct coauthors across the cataloged papers, with Muhammad Shahid Iqbal Malik the recurring name. Both papers are first-authored with no corresponding-author role recorded, an early pattern of lead-author writing within a small, stable circle rather than a broad network.
Activity—
Activity is unscored: the snapshot lacks per-year output counts and a continuous submission history. This axis would measure research cadence over time, such as outputs per year and posting continuity; the catalog currently shows only two 2026 preprints, too thin to establish a tempo.
Reproducibility78/100
Reproducibility scores 78, a mean over the 2 papers with reproducibility assessments. The SHAP/LIME rendering study rates high readiness, backed by OCR-measured error rates (CER 0.82-0.98) and a documented raster correction; the hate-speech framework rates medium, with inconsistent F1 reporting and leakage risk lowering confidence.
Research summary
Rameesha Zia is an early-stage researcher in natural language processing and explainable AI whose profile is name-resolved from public bibliographic data under ORCID 0009-0009-6717-2564; the available records list no institutional affiliation or country. The Evolve catalog holds two first-author arXiv preprints, both dated 2026, splitting exactly 0.5/0.5 between explainable-AI/NLP visualization and core NLP. The first, "When Explanations Cannot Be Read," measures how SHAP and LIME explanations render for right-to-left languages such as Urdu in Nastaliq script: Evolve judges it sound and well-controlled, reporting character-error rates of 0.82-0.98 for RTL labels and an OCR-measured raster correction that reaches the OCR ceiling, though word-level only and untested on humans. The second applies a DistilBERT-BiLSTM-attention hybrid with LIME to binary and multi-class hate speech detection; Evolve flags internally inconsistent F1 reporting, leakage risk, and missing significance tests. Reproducibility readiness averages 78 across the two papers (one rated high, one medium). The collaboration footprint is compact - 3 distinct coauthors, with Muhammad Shahid Iqbal Malik the recurring collaborator - and the collaboration dimension scores 28 while focus scores 0 on the even field split. Output, impact, and activity metrics are not yet available and remain unscored.
Collaboration pattern
The visible collaboration pattern is compact: 3 distinct coauthors across the cataloged papers, with Muhammad Shahid Iqbal Malik the only recurring external collaborator. Both works are first-authored, with no corresponding-author role recorded, consistent with an early-career researcher leading writing within a small stable team. With only two papers in scope, no co-authorship network structure can yet be inferred beyond this single repeated tie.
Research keywords
Highlights
Shows SHAP and LIME labels are effectively unreadable in right-to-left scripts - and fixes the rendering
First-author arXiv preprint (2026) measuring how SHAP/LIME explanations render for right-to-left languages such as Urdu in Nastaliq script: character-error rates of 0.82-0.98 make labels unreadable, and an OCR-measured raster correction reaches the OCR ceiling. Evolve rates it sound and well-controlled, though word-level only and untested on humans; reproducibility readiness is high.
Builds an explainable DistilBERT-BiLSTM-attention framework for hate speech detection
First-author arXiv preprint (2026) combining DistilBERT, a BiLSTM, and attention with LIME for binary and multi-class hate speech detection. Reported F1 scores are strong but internally inconsistent, with leakage risk and no significance tests, so Evolve treats the results cautiously and rates reproducibility readiness medium.
Timeline
- 2026Enters the Evolve catalog with two first-author arXiv preprints, both dated 2026 - the entire current record.
- 2026Measures SHAP/LIME rendering failures for right-to-left languages (Urdu, Nastaliq script), reporting character-error rates of 0.82-0.98 and an OCR-measured raster correction reaching the OCR ceiling.
- 2026Publishes a DistilBERT-BiLSTM-attention framework with LIME for binary and multi-class hate speech detection; Evolve flags leakage risk and inconsistent F1 reporting, rating reproducibility medium.
Papers in the Evolve catalog
Frequent co-authors
About this profile
This profile is AI-structured by Evolve.Pub from public bibliographic data (OpenAlex) and Evolve's own paper reviews. It is not written or endorsed by the researcher unless the profile is claimed. Radar dimensions are computed from real bibliometrics with fixed formulas — raw figures are shown alongside; missing dimensions mean insufficient data, never zero. Errors? Claim the profile to correct or hide it.
- Bibliometric scores are computed from the provided bibliometric snapshot (OpenAlex-derived) and cover only the 2 arXiv preprints (2026) in the Evolve catalog.
- No institutional affiliation or country is available; this profile is name-resolved from public bibliographic data via ORCID 0009-0009-6717-2564.
- Output, impact, and activity dimensions lack underlying data and are deliberately left unscored; no citation records exist for these recent preprints.
- The raw frequent-coauthor list contains the researcher's own name and a duplicated entry - apparent entity-resolution artifacts; Muhammad Shahid Iqbal Malik is treated as the sole recurring external collaborator.