Linguistic and Esthetic Testing of Kaspersky Lab Software
summary
Kaspersky Lab describes moving from slow manual checks to automated visual and text checks: they build AutoScreenshot and later Screenshot Web that run nightly, push diffs into TFS, and cut localization engineer time fivefold while halving agency linguist effort. The talk lists five error classes that break builds (translated variables, wrong escaping, encodings, duplicate hotkeys, inconsistent terms), shows metadata and pixel/ID delta comparison, and reports handling thousands of screenshots per run with usable bug reports.
AI-written summary · about
chapters (21)
- 0:00Introduction
- 3:28Localization process importance
- 7:05Costs of localization errors
- 10:25Manual to automated screenshot testing
- 14:07Automated linguistic quality tools
- 20:12Localization errors affecting functionality
- 23:36Internationalization best practices
- 28:33Summary
- 29:25Q&A: Automation & translation tools
- 30:15Q&A: Localization testing timing
- 31:20Q&A: Localization engineer skills
- 31:40Q&A: Problems found by automated tests
- 32:25Q&A: Text quality evaluation
- 33:34Q&A: Funniest localization cases
- 35:28Q&A: Labor costs & team involvement
- 36:52Q&A: Unknown languages in testing
- 37:37Q&A: Plugin development stages
- 38:22Q&A: Overnight screenshot testing
- 38:42Q&A: Machine translation services
- 39:09Q&A: Time allocation for testing
- 39:58Q&A: Localization data storage
related talks
27mContext: Missing ingredient in multilingual software translation
39mSpot the difference: automating visual regression testing
08m“Visual Testing”
37mThe Colorful World of Visual Test Automation
40mVisual regression testing: your strategy's blind spot
36mTest Automation to the Max
31mAutomating UX Compliance Testing
45mFrom Technical and Functional Testing to Linguistic and Cultural Testing