Some Thoughts on Work Records for Moving Images
Work records for moving images would be a high-value proposition because:
- Users often want to search on work-level attributes and we don't do a good job of encoding some of these attributes in our manifestation-level bibliographic records
- Moving images tend to be re-released in different variations (widescreen vs. fullscreen; director's cut vs. theatrical release) and format (DVD, VHS, etc.) so the ability to encode and reuse work-level data would have real economic benefits
- Videos are an important and popular part of many library collections (high circ in public libraries; used for films studies and lots of instructional purposes in academic libraries)
As far as I can tell, what we would need is something like the following:
- Infrastructure for creating, storing, and sharing work records
- Work records that contain:
- identifier for the work
- links to equivalent work identifiers (IMDB, allmovieguide, etc.)
- identifiers for or links to manifestations if these can be established (OCLC#, ISBN, publisher's numbers (non-standard, non-unique, basically messy, but usually in bib records), possibly some sort of search based on title and, at least for features, director)
- interesting information about the work in a useful form, such as:
- original title
- variant titles
- original language
- original release date
- country of production (most commonly defined as location of production company or companies)
- people (director, producer, writer, cast, etc.)
- production companies
- links to related works (based on, sequel to)
- summaries (possible problematic as most of the ones in existing bib records are cribbed from somewhere)
- subjects, genres, information about setting (place, time period, event), characters, etc.
- awards
- color, b&w, etc.
- sound or silent
- aspect ratio
- maybe user-contributed data, links to reviews, etc., etc.
- A mechanism for tracking where the info came from (extracted from bib records, IMBD/allmovieguide, title frames of video, cataloger's guess, publisher's website, reference book, etc.) as well as change history and possibly some sort of prioritization of reliability of sources.
It would also be desirable to provide an interface for searching, maybe something like OCLC's FictionFinder (http://fictionfinder.oclc.org) except with the ability to limit to the local collection.
It seems to me that a first pass at populating moving image work records could be made by extracting data from a large pool of bibliographic records or other sources followed by human review, particularly of records flagged as having contradictory data.
One approach would be to start with fiction feature films, which would probably a good-sized project because it's big enough to be interesting, but not so huge as to be overwhelming, and would provide better access to many popular materials through more consistent and computer-interpretable coding of characteristics such as original release date, original language, and genre. However, all types of moving image materials would benefit from at least a basic work-level record.
In the long run, it would also be helpful to enhance our current manifestation-level records with better encoding of distinguishing features.
In addition to what kind of data is useful and practical, there are also many technical and economic questions to be answered about how something like this might work.
Last updated: January 4, 2008
http://www.olacinc.org/capc/
Teressa Keenan