Hi fellas!
I may very well end up using Magma in a very interesting large scale project soon, and yes, I am also interested in querying capabilities that is being discussed in the other thread.
This is a related question: What free text search capabilities does Magma have today?
I admit I haven't looked, but wanted to ask anyway since I might get some info on the "plan".
And also - is this used/usable/interesting? http://map.squeak.org/packagebyname/fulltext
regards, Göran
The FullText package is a nice framework for churning through documents, updating them, and for doing some simple in-memory searches.
Combined with some of the more applicable index types in Magma, this would give rudimentary full text searching capabilities.
Missing: boolean operators, filtering out common words, scoring based on proximity of multiple search terms and/or frequency of occurrences, ...
So, if you need all that, it's maybe easier to dump your text into a MySQL full-text table (which worked surprisingly well the last time I used it - I have some VW code that interfaces with it), or into Lucene ("nobody gets fired for choosing Lucene"), or embrace-and-extent one of the FTX engines in C (Mnogosearch and Namazu come to mind).
Of course, if the project is sufficiently large scale enough, you can hire me to take care of this tasty morsel ;-)
On 12/20/05, goran@krampe.se goran@krampe.se wrote:
I may very well end up using Magma in a very interesting large scale project soon, and yes, I am also interested in querying capabilities that is being discussed in the other thread.
Hi!
Cees De Groot cdegroot@gmail.com wrote:
The FullText package is a nice framework for churning through documents, updating them, and for doing some simple in-memory searches.
Combined with some of the more applicable index types in Magma, this would give rudimentary full text searching capabilities.
Missing: boolean operators, filtering out common words, scoring based on proximity of multiple search terms and/or frequency of occurrences, ...
Mmm.
So, if you need all that, it's maybe easier to dump your text into a MySQL full-text table (which worked surprisingly well the last time I used it - I have some VW code that interfaces with it), or into Lucene ("nobody gets fired for choosing Lucene"), or embrace-and-extent one of the FTX engines in C (Mnogosearch and Namazu come to mind).
I found OpenFTS which sounds pretty good, based on PostgreSQL. But the problem is that yes - that would work fine for the server side solution - but I want also to do it locally on a lappy and a PGSQL installation is a bit over the top for that. :)
Of course, if the project is sufficiently large scale enough, you can hire me to take care of this tasty morsel ;-)
I bet. :) Well, we will see, who knows.
regards, Göran
On 12/20/05, goran@krampe.se goran@krampe.se wrote:
I found OpenFTS which sounds pretty good, based on PostgreSQL. But the problem is that yes - that would work fine for the server side solution
- but I want also to do it locally on a lappy and a PGSQL installation
is a bit over the top for that. :)
Namazu is probably simple enough to be embedded.
Of course, adding this to Magma would be a whole lot more interesting... Maybe time to collect some papers, algorithms, and stuff...
magma@lists.squeakfoundation.org