|
Work item:
|
F.DPLM-REM
|
|
Subject/title:
|
Requirements and evaluation methods of data provision capability for large models in multimedia applications
|
|
Status:
|
[Carried to next study period]
|
|
Approval process:
|
AAP
|
|
Type of work item:
|
Recommendation
|
|
Version:
|
New
|
|
Equivalent number:
|
-
|
|
Timing:
|
-
|
|
Liaison:
|
ITU-T SG13, ISO/IEC JTC 1/SC 42
|
|
Supporting members:
|
China Unicom, CAICT
|
|
Summary:
|
Data provision in multimedia applications refers to plan, organize, collect, process, store, distribute, and manage multimedia data to support large model training, which determines the performance of models.
Currently, there is a lack of unified and systematic professional standards in the development, management, and application of multimedia data. This has resulted in the need for large models to invest significant human and financial resources in tasks such as data cleaning when developing and utilizing multimedia data. As a consequence, it increases operational costs and severely impacts the industrial application of models. Therefore, evaluating the provision capability of multimedia data can assist both data application providers and data providers in understanding their own data service levels, providing guidance for future data development.
This draft new Recommendation specifies requirements and evaluation methods for data provision capability for large models in multimedia applications, so as to promote the data services for large models is controllable, manageable, traceable and auditable.
This draft recommendation aims to objectively evaluate the data provision level of industry participants, help product parties grasp the product level and provide optimization directions, help users quickly select and build data governance capabilities that meet the requirements, and promote the standardization and healthy development of the industry.
|
|
Comment:
|
-
|
|
Reference(s):
|
|
|
Historic references:
|
|
Contact(s):
|
|
| ITU-T A.5 justification(s): |
|
|
|
|
First registration in the WP:
2024-06-10 17:08:50
|
|
Last update:
2024-08-27 15:33:20
|
|
