Personal project

Geez Document Retrieval System

Dedicated Information Retrieval system for Ge'ez texts, featuring a custom stemmer and evaluated using precision, recall, and F1-score.

PythonNLPInformation Retrieval

What it does

  • Custom Ge'ez stemmer
  • Indexed Ge'ez corpus
  • Precision/Recall/F1 evaluation
Geez Document Retrieval System

Inside the build

Problem, constraints, decisions, and outcome.

What changed and why

01

Problem

Ge'ez texts are difficult to search because general-purpose retrieval tools do not account for the language's script and morphology.

02

Constraint

The work had to operate with a limited corpus and language resources while still producing measurable retrieval results.

03

My contribution

Built the retrieval pipeline, a custom Ge'ez stemmer, corpus indexing, search behavior, and evaluation workflow.

04

Technical decision

Use a dedicated information-retrieval pipeline and evaluate it directly with precision, recall, and F1 rather than relying on generic search behavior.

05

Outcome

Delivered a working Ge'ez search system with a custom linguistic preprocessing path and measurable retrieval performance.

More Work

Other projects from the same portfolio.