Trends in metadata

I am currently working on a Open Text Content Server (Livelink 9.7.1) project where one of the goals is to have standard attribute fields applied to all documents going into system. One of these attributes is "Document Type" (Policy, Procedure) which would contain all possible document types in use.

I can use tablekey lookup to narrow down this field to a subset of documents, e.g. first they click "Finance" which narrows the Document Type list to only Finance documents, but users do not like this. However, I also don't want them to wade through a long list of document types to find the one they want.

I'm wondering how other organizations have dealt with this. One example I have received was a Livelink customisation which autocompletes and filters the list on the fly as the user types.

Cheers,
Jeff

Tags: livelink, metadata, trend

▶ Reply to This

Replies to This Discussion

Permalink Reply by Billy Cripe on November 17, 2009 at 11:45am: Managing large metadata option lists is always a challenge. I prefer anything possible to be automated and/or suggested. For this reason I am a big fan of categorization, extraction and algorithm based assignments initially that are then supplemented with folksonomy (end user tagging) on an ongoing basis.

But here are other common strategies we use:

Dependent Choice List (aka DCL)- (select country which populates a list of states/provinces, select state/province which populates a list of cities, select city. Option to store the entire choice list or just the final value.)

Decision Tree - essentially a DCL but presented in a browesable mode like a windows explorer folder tree

Metadata Tag Clouds - Groups of associated metadata attributes clustered and sized according to prominence or weight, click on one and drill into a smaller cloud of terms associated with the clicked term

Recommendation Engine - as docs go into the system we can run them through our "categorizer" application which "reads" the doc and recommends metadata for different fields. Recommendations are based on a heuristic that the admin defines so it can be complex or simple.

Automated assignment rules - based on strict or known criteria (if-then) pre-populate the metadata fields with some value. For example auto populate a file type metadata field based on file extension.

Workflow assignment rules - get the doc into the system with minimal metadata attribution and then run it through a workflow process and allow humans to categorize/attribute it.; ▶ Reply to This