About CAVA
Conversation is the means by which we manage our identity and relationships, and influence the world around us
Background
The objective of the CAVA project is to establish a repository for audio-visual data on real-life human communication for spoken and signed languages. In order to investigate human communication and interaction, researchers need hours of audio-visual data, sometimes recorded over periods of months or years. The process of collecting, cataloguing and transcribing such valuable data is time-consuming and expensive. Once it is collected and ready to use, it makes sense to get the maximum value from it by reusing it and sharing it among the research community. Historically, the study of communication has been based on highly-controlled experimental data, but a better understanding comes from examining natural audio-visual data. Such work, both qualitative and quantitative, involves in-depth study of video- and audio-recorded data of conversations and clinical encounters.
But unlike highly-controlled experimental data, natural audio-visual data tends to defy easy classification, and may lead to idiosyncratic solutions to preservation, metadata and access issues. It is not uncommon for vital and unique data to languish on VHS tapes in personal collections; researchers across the discipline waste time battling with increasingly inaccessible media and finding individual solutions to the challenges of editing and analysis. The resources of funders, researchers and subjects are wasted on the collection of new data rather than on the re-use of existing data. Natural data can often be used for more than the purpose its collector intended. Researchers may be able to save time and money, or improve the depth of their observations and conclusions, by reusing existing data instead of collecting their own.
Despite its usefulness, data in personal collections does not lend itself to being shared between researchers and institutions on a large scale. This is largely due to the absence, until now, of a centralised data archive to support such research and to offer opportunities for collaborative work.
History and aims
CAVA was the product of a UCL Research Challenges grant which began in November 2007. This allowed the team to investigate the feasibility of centrally archiving data held by the Centre for Applied Interaction Research (CAIR), an interdisciplinary grouping largely based in the UCL Division of Psychology and Language Sciences. The CAIR project investigated a discipline-specific metadata standard, and archived a pilot sample of data for dissemination through UCL's Moodle virtual learning environment. A feasibility study conducted as part of the CAIR project also found considerable support among the research community for a comprehensive and accessible repository.
CAVA will create a repository of re-usable video material to support the work of UCL and of the international human communication research community. The UCL team already holds a large body of content (over 600 hours of material), in suitable formats, with appropriate permissions secured. The data mainly comprises videotaped interactions - conversations, interviews, assessments - between a person who has atypical communication (due to disabilities such as stroke, deafness, autism etc) and their spouse, teacher, parent or another 'typical' communicator, filmed in the home, at school or in the clinic. The duration of the videos ranges from 10 minutes to more than an hour per client. Some of the data is longitudinal with regular sessions filmed over a period of time.
Preservation
An important goal of CAVA is to centralise data held in personal collections and make it easily searchable by uniting the disparate sets of information held by the original researcher. The initial dataset will come from CAIR members based at UCL, though researchers at other institutions have expressed interest in contributing. The majority of this data is stored on portable media or cassettes, which run using proprietary or hard-to-find codecs, meaning that they are difficult to share and are at risk of obsolescence. The data will be captured from its original medium, whether CD, VHS or Mini-DV cassette, and transcoded into low-compression AVI format for preservation. The preservation file will maintain the quality of the original to the highest possible extent. This represents an acceptable compromise between fully uncompressed, but unwieldy, preservation files and the risk of the original formats becoming obsolete. The objective at this point is to achieve uniformity of data in order to aid long-term management of the dataset.
Discovery
It is not enough, though, just to collect and standardise the quality of the data; it must be readily searchable. CAVA uses a modified metadata standard based on the ISLE MetaData Initiative (IMDI), a schema designed for language resources. The nature of the data presents some crucial challenges to the creation of metadata. Implementing the full IMDI standard would be too time-consuming and costly, for both the project team and depositors. The key issue to address is that of multiple participants. Based on various modifications of the IMDI standard, principally the UCL Deafness, Cognition and Language Research unit subset, the CAVA subset presents a pragmatic solution. A mapping between UCL's IMDI instance and the Dublin Core standard will be written. A full metadata schema and best-practice guides for capturing data (for both users and potential depositors) wil be placed on this website.
Storage and dissemination
As the data is collected it will be stored using the UCL Library Services Digital Collections service. The team will devise and test ingest processes so that video clips, transcripts (where available) and descriptive metadata can be uploaded to the repository in batches, in a way which maintains the relationships between the one or more versions of each video recording, its transcript, and the metadata which applies to each. The final ingest process will include the automatic generation of technical metadata and the creation of appropriate access restrictions.
The data will be made available in several dissemination formats. All data accepted by the archive will have appropriate permissions for the various types of dissemination. Users will be available to download compressed video or uncompressed audio-only files. All dissemination formats will be prepared in order to operate with minimal codec and system requirements.
Access management
In order for a researcher to benefit from access to the data, they must be able to manipulate the files at their leisure. However, in order to encourage researchers to use the archive, and primarily to request access, the dissemination will operate on a tiered basis. Our key concerns at this stage are to ensure good procedures for data protection, identity and ethical issues. The metadata will be searchable through the DigiTool front page, although none of the data itself will be viewable at this point. A login will be supplied to researchers to permit them to view streamed excerpts of the data which potentially interest them. The researcher would then request full access to downloadable versions of their selected datasets. By these means, CAVA takes all reasonable precautions to prevent the often-sensitive data from being used inappropriately.
The project team will work with the UKDA to design the application process for prospective users, implement procedures to verify and authorise requests, and register and authenticate users. The retention and presentation of rights information will be implemented within the IMDI record. Users will see hard-copy or click-through licensing agreements to be associated with particular tiers of access, to indicate clearly and unambiguously what an authorised user may and may not do with the material.
July 2009
