# Context Evaluation Report

> Evaluate context sufficiency to help improve translation accuracy

URL: https://store.crowdin.com/context-evaluation-report
Identifier: context-evaluation-report
Category: AI, Manager Productivity
Type: App
Author: Crowdin
OpenAPI spec: https://store.crowdin.com/context-evaluation-report/openapi.json

UI localization projects are typically a set of short texts that are hard to translate because they might have several valid translations but only one or a few that fit the actual usage context. The shorter the text is, and the more target languages you have - the bigger the problem becomes.

Crowdin provides a set of tools to address this issue. Ideally, UI strings should be accompanied by [screenshots](https://support.crowdin.com/developer/automating-screenshot-management/), giving linguists a very clear idea of where each text is used without interrupting their translation workflow. Another approach is using [In-Context](https://support.crowdin.com/developer/in-context-localization/), which yields the best translation quality but may break a linguist's workflow because they need to learn the software they localize. There are also automated or [AI context extractors](https://github.com/crowdin/context-harvester) that analyze the source code for clues. Even more approaches exist, like the [Figma integration](https://support.crowdin.com/figma-plugin/#uploading-tagged-screenshots-to-crowdin) to supply context or letting linguists request additional details if they find a segment that can't be translated without more input.

## How This App Helps

This app provides one more way to address the context problem: an AI-powered **Context Evaluation Report**. This app helps predict which of your segments are at risk of getting poor translations because of missing or insufficient context.

When installed, it registers a new report in Crowdin – "Context Evaluation Report". You can select the scope of what's being checked (e.g., per file or for tasks after they're created but before they're assigned) and optionally set a timeframe to limit the report to recently added strings.

Once the report is generated, any keys with potential context issues get highlighted. You can export the report (share it with your team if needed), fix the problems (e.g., add extra notes), and then upload the updated report back to Crowdin so linguists have the improved context.

## Prompt Configuration

After installing, the app also creates an AI prompt you can use as a template. You'll need to finish configuring it (add an AI provider and model). For the best results, we recommend choosing AI models with reasoning capability.

## API

If you want to automate the report process, you can make requests to Crowdin's platform API, which securely routes calls to this app on your behalf. The following actions are supported:

- **Generate a Context Evaluation Report**  
- **Check Report Generation Status**  
- **Cancel an Ongoing Report**  
- **Download a Completed Report**  

For full details and request examples, see the [Context Evaluation Report API docs](https://context-evaluation-report.app.crowdin.net/api-docs).

## Screenshot
![Context Quality report in Crowdin](https://developers.app.crowdin.net/api/media/file/1371aa7a-4918-4f68-9242-70cd1280deea.png)

## Support

If you have any questions or need help setting up the Context Evaluation Report, please contact our support team. We’ll be happy to assist!

## Resources

- [API Docs](https://context-evaluation-report.app.crowdin.net/api-docs): [OpenAPI spec](https://context-evaluation-report.app.crowdin.net/api-docs/swagger.json)