On Sunday 28 Jul 2013 10:29:25 leuchtkaefer wrote: > This is an update of the current state of plugin Curator including future > plans. > > Sharing content using Curator: > > Content curation is the process of collecting and classifying the most > interesting things about a subject to share it for the common benefit at a > single location. WoT Identities are used to sign the curated material; thus > each identity owns and maintains a unique index. > > The plugin depends on WebOfTrust and Library. In a nutshell, WebOfTrust > provides the available own identities. It also announces which Identities > publish an index and the location of such index. Library creates the index, > handles an index on-disk version, uploads the index to Freenet and merges > index when is necessary. > > Currently, a big limitation in Library is that it can manage only one index > for Spider. My current task is to adapt Library to manage multiple indexes. > For that task the main important classes are > plugins/Library/SpiderIndexUploader.java and > plugins/Library/SpiderIndexURIs.java. At the moment, I am working on a copy: > CuratorIndexUploader.java and CuratorIndexURIs.java. Both are being modified > to handle multiple ProtoIndex, merge new data and do the uploads according to > different indexes. Indexes are identified by the routing key part of the WOT > identity RequestURI.
Ok. > > At the moment Library still handles only one index since the modifications > are not yet finished. I decided to publish the plugin at current state while > I continue to the adaptation of Library. The code can be found in a Github > repository (https://github.com/leuchtkaefer/contentcuration) I will review this soon. > > All entries in the index need to be SSK keys. The reason for this is that I > am using the TermEntryWriter that stores data in binary using the method > FreenetURI.writeFullBinaryKeyWithLength(DataOutputStream dos). The FreeenURI > method doesn't handle USK keys, I don't know why. This is a deliberate design decision. USKs are *updatable*, meaning if the index points to a USK it's probably wrong: we convert it to an SSK so it points to a specific edition. When we show the search results we can convert it back to a USK, but we need to show the SSK link as well, since we know that *that edition* contained the content we were interested in (later editions may not). > > To use the plugin, Library needs to be updated since I did some modifications > on that plugin too. If you need to run Curator, please tell me where I can > commit my Library changes. Fork Library on github into your own Library repository, and push changes there. Please do not "drop a bomb" (a large amount of code in a single diff or single commit) - use git properly. > > Finally, I want to mention next work: > - finish adaptation on Library to handle multiple indexes Right. > - add bookmarklet button so the user can Indexes will further share with > other identities who may I do not understand this line. > - have a list of terms showed by the input form, so the user can select the > topic (category) easily. I do not follow - you want to manually label the files with keywords? This is a good idea, but if they are multi-word there may be complications. Initially we can just index by words extracted from the description or filename? > - Ideas for future work on the plugin (cannot estimate the amount of work > needed): > - search terms under indexes (search from trusted identities, use some > threshold) Right. This is key. > - reuse indexes curated by other identities to generate new indexes. This is future stuff. > > Leuchtkaefer
signature.asc
Description: This is a digitally signed message part.
_______________________________________________ Devl mailing list Devl@freenetproject.org https://emu.freenetproject.org/cgi-bin/mailman/listinfo/devl
