contact@icresg.org
From Ideas to Impact
ICResG From Ideas to Impact
Join ICResG

Natural Language Processing

Multilingual Personality Computing for Low-Resource South Asian Languages

Develop ethical, interpretable and culturally aware personality-computing resources and models for low-resource South Asian languages, beginning with Urdu and Kashmiri where feasible and extending only through qualified language collaborators.

0
Progress
2
Researchers

About this project

This project will study how personality-related linguistic signals transfer across languages, scripts and cultural contexts that are under-represented in current NLP research. The team will design an ethics-first protocol, identify appropriately licensed or consented text resources, develop transparent annotation guidance, and compare classical, transformer and graph-based multilingual models. Outputs will include a documented benchmark or feasibility dataset where permissions allow, reproducible baselines, fairness and cross-cultural validity analysis, and a research paper. The project will not infer personality for surveillance, employment screening or other high-stakes decisions.

Objectives

  • Develop an ethics and governance protocol for multilingual personality-computing research
  • Assess available Urdu and Kashmiri resources, licences and data gaps
  • Create transparent annotation and quality-control guidance
  • Compare linguistic, transformer and graph-based multilingual representations
  • Evaluate cross-lingual transfer, calibration, bias and explanation quality
  • Release reproducible methods and publish the findings

Research priorities

  • Low-resource and multilingual NLP
  • Personality computing and computational psycholinguistics
  • Graph representation learning
  • Cross-cultural validity and fairness
  • Explainable and responsible AI
  • Reproducible language-resource development

Sustainable Development Goals

  • SDG 4 · Quality education
  • SDG 9 · Industry, innovation and infrastructure
  • SDG 10 · Reduced inequalities
  • SDG 16 · Peace, justice and strong institutions

Activities

  • Protocol and ethics review
  • Language-resource and literature mapping
  • Data provenance, consent and licensing audit
  • Annotation guideline development and pilot study
  • Baseline, transformer and graph-model experiments
  • Bias, calibration, robustness and explainability evaluation
  • Dataset documentation, code and manuscript preparation

How to take part

Language experts, NLP researchers, psychologists, responsible-AI researchers and data-quality contributors can join defined work packages. Native or advanced language competence must be documented for language-specific annotation, and all contributions will be tracked for contribution-based authorship.

Skills required

  • Python
  • NLP
  • multilingual modelling
  • transformers
  • graph learning
  • annotation
  • statistics
  • research ethics
  • scholarly writing

Roles needed

  • NLP researcher
  • Urdu or Kashmiri language expert
  • Graph learning researcher
  • Responsible AI researcher
  • Data and evaluation researcher

Who should apply

Open to postgraduate students and researchers with relevant NLP, language, psychology, graph-learning, evaluation or responsible-AI skills. Language-specific contributors must have demonstrable competence in the relevant language.

Expected contribution

Collaborators must use ethically sourced data, respect licensing and consent boundaries, document methods, participate in regular meetings, use reproducible version-controlled workflows, and contribute substantively to assigned outputs.

Related work

Other projects in Natural Language Processing.

All projects

Newsletter

Keep up with the group.

Calls for collaborators, papers and events — a short note when there is something worth saying.

The address is not shared, and one click leaves the list.