An exhaustive, representative test database in phrase search algorithm

Question

For a phrase searching algorithm, imagine the goal is to search for a name phrase and return matched results based on a pre-defined threshold. For example, searching for "Jon Smith" could return "Jon Smith", "Jonathan Smith", "Jonathan David Smith", "Jonathan Smith-Mikel", "Jonathan 'Smith' Mikel" etc.

The plan is to manually choose N test cases and put them in a benchmark database. I have concern about this plan because the test cases is likely to be not exhaustive. I know there are pretty mature search engines there, so is there a test database which covers all possible cases, such as different name combinations, punctuations, symbols, etc. such that we can use this as our benchmarks instead of guessing?

For example, this test database should contain all cases for "Jon Smith", as well as connectors such as hyphen, apostrophe, and so on.

Answer 1

Based on the example I assume that the target is persons names.

Let's be clear, there's no such thing as an exhaustive dataset containing all possible persons names in the world. Also a crucial part of the question is: in What kind of names? In which language? Persons names in English are pretty different from names in Chinese for instance. And there is also the difficult question of transliteration of proper names.

That being said, a few resources exist. They would usually be found by searching resources for "personal names", "record linkage", "named entities matching/coresolution". The following ones probably don't cover all the requirements but it's a start:

• Febrl (see also here)
• Found This paper which presents a large resource and this corresponding resource description but couldn't find the data.
• The EMM news explorer has an interesting database of named entities including persons names with all the spelling variants/transliterations.