The Nepomuk development is split across many different repositories across KDE. The page attempts to list out most of the known Nepomuk code, and its development location. The list has been split according to importance.
These are the official repositories which the core Nepomuk team focuses on.
All Nepomuk enabled applications must depend on NepomukCore. It utilizes the Nepomuk2 namespace.
NepomukWidget, is the repository which contains all the widgets which applications use. It will be released with KDE 4.10. Before this the widgets were present in kdelibs/nepomuk.
Nepomuk currently consists of 3 kio-slaves - nepomuk, nepomuksearch, and timeline. Out of these 3, nepomuksearch and timeline are the most visible. The nepomuk kioslave is an internal kioslave which redirects to the physical location or to the nepomuksearch kioslave. For a more in depth discussion of their working, please read this.
The Strigi project provides libraries to extract metadata from file. The project is divided into 5 sub-proejcts. Nepomuk only requires libstreams and libstreamanalyzer.
The overall development of Strigi has been stagnant for a couple of years now. However there are no viable alternatives present. And the cost of moving would be risking stability.
Strigi requires a lot of work where all the file formats can be listed, and the compatibility for every file can be checked.
From KDE/4.7, Nepomuk has started enforcing the ontologies - Either the data follows the ontologies or it is rejected. Since this move many bugs and inconsistencies have come to light. Unfortunately, while earlier they were just an minor inconvenience to the user, now those files are not indexed.
The Nepomuk web-extractor was originally a 2010 GSOC Project (Add link). The project was completed, but the architecture is extremely complicated. Additionally, most of the user visible parts were missing. It requires a complete rewrite.
The Nepomuk Metadata Extractor is an independent project that was started by Jörg Ehrichs. It provides rich features for extracting additional information from the web as well as from the files.
A fairly random project that was done in less than a day. Its purpose was to provide a user visible way to rename, delete and merge tags. It is currently not shipped by any distributions, and should ideally be cleanup up and properly released. For more information read this.