Rethinking Big Data Testing: Defining Problem Domains and Key Test Areas

Internet growth gave rise to the big data industry; big data platforms and applications have become critical testing targets.

Viewing the problem from the problem‑domain perspective

A problem domain is the set of related issues and logical space surrounding a question. In testing we usually start with categories like functional, performance, or stress testing, then map concrete questions to a domain.

Example questions on a search page

What result does searching “Jay Chou” return? Can it provide his phone number?

What happens if the search box is left empty and the button is clicked?

What is the outcome of clicking the camera icon and searching an unknown flower?

Is the recommendation aligned with user preferences?

Is the hot‑search list real‑time?

How to view Shanghai weather?

Is voice playback normal?

What happens when gibberish is entered?

These can be grouped into functional testing, but for big‑data platforms we can further split them into more specific domains.

Big Data Platform Problem Domain

Data extraction verification – ensure correct and efficient extraction from diverse sources (text, image, audio). Example: verify text and image search in an e‑commerce platform.

Data conversion and transmission verification – check correctness when data is transformed and moved between components.

Data loading and display verification – confirm that loaded data is presented properly to users (layout, image clarity, device adaptation).

Big Data Application Problem Domain

Real‑time data update testing – validate that newly generated data appears promptly in the application (e.g., search results update instantly during a flash sale).

Algorithm effectiveness verification – test recommendation or ranking algorithms, such as limit‑flow checks for content publishing.

Response timeliness – measure latency of user‑facing operations, crucial for high‑traffic scenarios.

Algorithm stability – assess consistency of algorithm behavior over time and across environments.

Conclusion

Instead of starting from traditional test categories, enumerate many concrete questions, classify them into problem domains, and then expand each domain into focused test activities. This perspective can inspire more systematic big‑data testing strategies.