Methodology
Content Processing
In order to put voice back into oral history, the audio and/or video media must be easily accessible whereby the user can readily browse passages on a particular topic.
a database of indexed topics must be developed for each oral history collection.
Using an off-the-shelf audio/video indexing database software (InterClipper™), the interviews are assembled, with the media linked directly to database records for annotation and metadata development.
OHIAP content processing methods were broadly structured by:
our goal of providing a searchable database to the public, students (k-12 and college), and professionals via the WWW
the resources required to process over xxx hours of interviews.
media file size, knowledge level of database user, and ease of finding both broad conceptual topics and specific things related to Illinois farm life and agriculture were considered
Our content processing methods have tried to strike a balance between not only widely variable end-user knowledge/interest levels, but also with the level of topic specificity theoretically possible and that which is practical.
Steps to Create Searchable Interview Content Database Using InterClipper
Divide Interviews (wav or avi file) Into 8-to-10 Minute Segments (SEGS)
Each SEG may contain a discussion of one or more topics
break points between SEGS should not be placed such that the narrator’s discussion of a particular topic appears in 2 adjacent SEGS
screenshot here
Compose paraphrased annotations for each SEG using thematic terms (CONTROL WORDS), and key names, places, and dates (NAMED THINGS)
provide a comprehensive but reasonably "bite-sized" access to all of the interview material
The words will be used to search the data to isolate specific passages based on a theme or combination of themes
screenshot here
Highlight STORY-CLIPS: In each SEG, selected passages are defined and annotated (again using CONTROL WORDS and NAMED THINGS)
STORY-CLIPS are short (1-2 minute, but usually less than a minute) passages of interest marking more particular, bounded stories, anecdotes, statements, and interchanges of interest
STORY-CLIPS must be engaging and entice the visitor to listen to more than one STORY-CLIP as well as explore the database
screenshot here
CONTROL WORD Development
Initial annotations are the basis for developing a list of Control Words, which is iteratively revised and updated with growing number of interviews processed.
The words will be used to search the data to isolate specific passages based on a theme or combination of themes.
To ensure a user can find a specific item of interest, e.g., John Deere tractor, Sangamon Co., IL, chicken, a list of NAMED THINGS is developed concurrently.
screenshot & CONWORD LIST
graphic example showing reduction in number of SEG & STORY-CLIP results with use of several CONWORDS here
Data Migration
Migration of the InterClipper OHAIP database to Microsoft Access and MySQL is necessary to serve the OHIAP database on the web and facilitate efficient web-based queries. This migration will allow professionals to import the OHIAP database into other database software for off-line use.
screenshot here
Results
To date all 18 interviews in the NIU Oral History Collection, 44 interviews in the UIS Oral History Collection, and 6 new interviews in the ISM Oral History Collection have been indexed for content. This represents a total of XXX hours of interview, XXX segments, and XXX story-clips.
Although CONTROL WORD development and refinement continues, it is sufficiently robust to provide examples illustrating the power of the indexing method.