Explore grouping of search results under the same document

[cjcenizal]

Problem

As demonstrated by the first and third results in this screenshot, occasionally results can appear as "duplicates" because a single document can contain multiple relevant references, each of which counts as a result.

Solution

From a UX perspective, one solution could be to visually group chunks together if they come from the same document, so there won’t be any repetition of documents. The tradeoff is that the results listed at the top have a higher relevance score than those below, so I wonder if it will be confusing if a chunk from the 5th result (less relevant) is grouped with a chunk from the 1st result (more relevant)?

I also wonder if multiple references from the same document carries enough significance to change how we present the document to the user. For example, should we present it as some sort of "primary" resource? I doubt we can make that kind of assertion, without knowing more about the nature of the documents being used. And anyway, having multiple references grouped under a single result will naturally lend it more visual weight.

References

Here's how other folks have solved this kind of problem.

Perplexity