Evaluating and redesigning the OCR workflow to improve accuracy and multilingual formatting.
The Newar Online Dictionary team needed a more reliable way to convert their multilingual dictionary manuscripts — written in Devanagari and Latin scripts — into accurate, editable digital text. Their existing OCR workflow produced inconsistent results and required extensive manual correction.
Revitalization Technology partnered with the existing team of talented developers and dictionary editors to evaluate their options and design a more sustainable and accurate solution.
Our work focused on three key areas:
Evaluating OCR Tools for Devanagari Script
We conducted a comparative review of OCR engines capable of handling Devanagari and mixed‑script documents. This included assessing accuracy on real samples from the project, testing how each tool handled formatting, and identifying where additional training or customization might be needed.
Assessing Cost, Licensing, and Data Security
Because the team works with sensitive linguistic materials and large volumes of scanned pages, we analyzed each tool’s pricing model, data‑handling practices, and long‑term sustainability. This ensured the final recommendation aligned with both the project’s budget and its commitment to community data stewardship.
Identifying the Best Fit for Multilingual Dictionary Formatting
The NOD manuscripts include complex formatting including headwords, example sentences, multiple languages, and structured lexical information. We tested how each OCR option preserved or distorted this structure and identified the tool that produced the cleanest, most consistent output for downstream editing.
The Result
Through this process, Revitalization Technology delivered a clear recommendation and implementation plan that improved accuracy, reduced manual cleanup, and strengthened the project’s long‑term technical foundation. The result was an OCR workflow that better supports the team’s ongoing digitization efforts and aligns with their goals for sustainable community dictionary development.
