FTS Search and normal search should help each other and not have their own world

Today if you search with FTS only, then it will search only via the Full Text Search engine.

If you search via the search criteria (without using the FTS search), then the search will search through the metadata of CMOD

Now, the problem is, when you mix the 2 searches:

For example, I search for all documents between 2 dates, then I give a document type, and a customer number. This is done through the normal search of CMOD (no FTS). I might get something like 30-40 documents, instead of 100000 documents.

Now in these 30-40 documents, I know I want to see all documents referencing “IBM Switzerland”, which is amounts maybe for 3 or 4 documents. So, I add the FTS search to search for these terms “IBM Switzerland”.

The result TODAY is that you need to wait many seconds, even some minutes (until the timeout) until you get the answer.

If in the FTS Search field, I add the customer number, in addition to “IBM Switzerland”, then I get the answer in a matter of seconds, if not milliseconds.

In this example, we can see that you are already reducing the result with the normal search, you need to duplicate some metadata in the normal search, to add them in the FTS search.

No user wants to do that (I mean duplicates the same values in 2-3 fields). The user expects that in the 30-40 documents, and only in these ones, you do the FTS search. And the answer should be lightning fast.

That’s why the idea here, is really that the 2 searches work together in order to find the documents that the user wants in a timely manner.

The user should NOT need to learn that the 2 searches are separate and works completely separated from one another.

This will help CMOD to have a much better user experience.

How should it work?

Here are a few ideas, that I have in mind:

1. every value put in the normal search, are automagically duplicated in the FTS search field, behind the scene, ONLY IF the FTS search field is filled by the user. From what I’ve heard when I’ve opened some FTS Ticket by IBM, then the FTS engine also store the indexes of the documents.

2. Enhance the FTS Engine to accept to search on a subset of the documents, I mean if the normal search returns 30-40 documents, then in addition to the search from the FTS, you pass the documents found with the FTS, and then the FTS Engine does its magic only on those 30-40 documents.

Because our users were really not happy, we have implemented a workaround through a plugin in ICN (Content Navigator) to implement the solution 1). And now the users are happy, because they get a FAST answer.

The problem is that we can do it with ICN, but we cannot do it with the Windows Client, and some of our users are still not happy, but they are using a bit more ICN.