Hi,

I have recently started using Hibernate Search and have found it Great!!!

I have read at a number of places including "HS in Action" that the results returned are sorted by Relevance. What does this mean? Does this mean that if I am searching through names for "John", then the results would be sorted as follows.

John A
John B
John C
A John
B John
C John
A A John
A B John

and so on.

I hope I am making myself clear. The results are sorted not only based on the position of the search term but also by last name(where applicable) or first name (where applicable)...

I tried a simple example, but the results were not as above, maybe I am making some mistake.

If this is not the default behavior, is it possible to achieve this using Hibernate Search?

Any help would be highly appreciated.

Regards,
Shardul.

Hi,
well one of the coolest thing is that unlike DB-level proprietary full text engines you can tune your own notion of what "Relevance" should mean with quite a lot of flexibility. The default relevante is very nice, but in this applications you want to be able to hint the system about what you think your particular use case should return, what your users are expecting to find.

The default "Relevance" concept is not about the position of the terms being at the beginning of the text nor about lexicographical ordering, it's about the terms you are searching for are more frequently used in one text, how good they match, how far they are from each other (supposing larger natural text sentences); in case of more than one search term, you will get first the results containing most frequently both terms, and even a better score if the two terms are used near each other.

In the book you really want to take a look to chapter 12: scoring and boosting of results. It has code examples and lots of good explanations about how it's done and how to best tune it to your needs.

Thanks for your inputs....am going through chapter 12, would get back in case I have any questions.