Posted by benoit.borrel on January 18, 2010 at 9:25pm
Last updated by benoit.borrel on Fri, 2010-10-08 00:14
Last updated by benoit.borrel on Fri, 2010-10-08 00:14
Many modules offer, based upon different methods, functionalities that serve to detect relation of similarity between nodes (sometimes named "more like this", relevant, similar...). By methods I mean modules relying on any form of automatic (or semi-automatic) process, which detects and lists similar nodes. Modules that solely offers a manual-based listing process are therefore out of scope.
Below is a proposition of classification of methods I would like to discuss with the community:
- taxonomy/CCK based, which compute a match score.
- simple terms matching (ex: Similar by terms)
- complex/custom defined weight and compound associations (ex: Associated nodes)
- content based
- MySQL full text searching, using natural language (ex: Similar entries)
- semantic similarity, like the methods listed in semantic relatedness (but as far as I know no modules implement any of them) - as of October 5th, 2010, I published Semantic Similarity module.
Note:
* automatic tagging modules may help streamlining the whole process of nodes' relations detection when taxonomy based.
* some other interesting pages are Comparison of Similiar / Relevant by term block modules and Related content modules.