Member of the PREMIS Task Force

Transcription

1 Using PREMIS to support preservation of digital assets at Yale Yale University Integrated Access Council [2 nd ] Preservation Metadata Task Force 2006 October 2006 Member of the PREMIS Task Force Matthew Beacom, Metadata Librarian, Library Catalog Dept. (Co-chair) Reed Beaman, Associate Director for Biodiversity Informatics, Peabody Museum Lee Faulkner, Media Director, Digital Media Center for the Arts David Gewirtz, Project Manager, Library Projects, ITS Kevin Glick, Electronic Records Archivist, Library Manuscripts and Archives Rebekah Irwin, Catalog Librarian for Digital Projects, Beinecke Library (Co-chair) Edward Kairiss, Director, Instructional Computing Instructional Technology, ITS Daniel Lee, E-Publishing/Internet Marketing Manager, Yale University Press Youn Noh, Digital Resources Catalog Librarian, Library Catalog Dept. George Ouellette, Senior Programmer Analyst, Library ILTS Thomas Raich, Associate Director, Information Technology, Art Gallery David Walls, Preservation Librarian, Library Preservation Dept. Background In March 2006, the Yale University Library Integrated Access Council accepted the recommendation of its first Preservation Metadata Task Force to adopt the PREMIS model as the basis for preservation metadata for its digital information assets. 1 As a practical next step toward implementation of PREMIS-based preservation metadata, a second task force was charged in April 2006 to investigate, develop, and propose a metadata element set and usage guidelines based on the PREMIS model. The scope of PREMIS is limited to metadata that supports preservation activities. PREMIS-based metadata is only one part of a larger package of metadata needed to support the use and re-use of digital assets held within a repository. Other metadata standards complement PREMIS. Additional metadata for digital information objects would include, for example, MARCXML or MODS for descriptive metadata that supports discovery of information objects, NISO MIX for technical metadata that describes the physical characteristics of images, and METS to contain or wrap together combinations of metadata types for transmission between applications or repositories. PREMIS is part of a set of metadata tools with which Yale can manage its digital assets, provide for retention and re-use for the long term, and build information services for the Yale community and the world. In our discussions and analysis of using PREMIS at Yale, two of our most critical concerns have been effectiveness and cost. To be of value, the proposed element set and guidelines must both support the preservation of digital information assets at Yale and meet the practical needs of digital asset administrators. Administrators need to weigh the cost and effectiveness of generating and storing additional technical and 1 Recommendation to Adopt PREMIS as a Preservation Metadata Model for the Yale University Library (1/12/06) [PDF, 11 p.] Yale University / Preservation Metadata Task Force Using PREMIS to support preservation of digital assets at Yale / 1-1

2 preservation metadata, taking into account the stability of the format, the value of the object, and obligations to third parties. PREMIS is designed to be an effective and inexpensive implementable tool that provides the metadata or information needed to preserve digital information assets for the long term. 2 To that end, the PREMIS elements are designed to be information that is collected through automated or semi-automated processes. No PREMIS element should require manual input per digital object. Although they may not be fully automated across every application, they can certainly be semiautomated. The process to create or capture preservation metadata should be economically sustainable. Costs also may arise from developing the automated tools needed to create or capture preservation metadata and the policies to support their use. For example, the base profile has only a tiny number of elements that are required, but each element requires work to develop tools and policies. Yale would need to do this development quickly and efficiently, or the result could be a protracted development phase. Measuring the effectiveness of PREMIS metadata needs to be based on experience with the performance of the metadata in actual instances of use. Preservation of digital information assets is new. Trials using PREMIS are now underway around the world. Two of our recommendations below address the need to engage with peers implementing and evaluating PREMIS while beginning to use PREMIS locally in an experimental evaluative spirit. There is, as well, a third related and critical issue that must be considered with effectiveness and cost. That concern may be thought of as relevance. In short, all digital assets at Yale may not warrant an application of metadata suitable for longterm preservation. Although administrators of digital asset collections need guidelines for application and use of preservation metadata at Yale, these guidelines must be relevant to a wide range of digital asset collections. For example, administrators of e- records in a university archive may have a different obligation for long term preservation than do administrators of digital surrogates of items in museum and library collections, or administrators of e-reserve course materials. The range of digital asset collections at Yale requires an adaptable framework for determining the appropriate degree of preservation responsibility per collection, repository, or administrative unit. PREMIS cannot make policy decisions. PREMIS cannot provide guidance on appropriate levels of responsibility for preservation of Yale s digital assets. However, PREMIS can supply a critical piece of the digital preservation infrastructure at Yale. It is a building block well-suited to Yale s wide ranging digital asset collections and systems implementations. What is needed is a larger policy framework for applying the PREMIS model at Yale. 2 Data Dictionary for Preservation Metadata: Final Report of the PREMIS Working Group [PDF:3.2MB / 237p.] Yale University / Preservation Metadata Task Force Using PREMIS to support preservation of digital assets at Yale / 1-2

3 Recognizing this diversity of needs among digital asset administrators, we understand that there is no one size fits all approach to preservation metadata. The library and its peers at Yale need to develop policies to determine when to apply PREMIS. The task force recommends that Yale develop a small set of preservation metadata profiles for distinct yet relatively generic levels of obligation for preservation of digital assets. The profiles would guide appropriate use of PREMIS across a continuum of obligation. administrators would review semantic units in PREMIS to insure that they are capturing the information needed to extend a base profile or abbreviate a full profile to their domain-specific profiles. Between the extremes of a base profile or a full profile, the library and its peers at Yale need to develop a set of profiles that relate preservation policies with appropriate metadata requirements. Development of such profiles will depend on a number of policy decisions made by various organizational units at Yale, such as the library, the gallery, natural history museum, and ITS. The staff developing these profiles should be experienced in preservation activities, digital information technology, and metadata services. These profiles should be generic or neutral with respect to specific information systems implementation. Two levels are immediately apparent, a base level and a full level. Two Profiles Base profile. This profile would provide guidance to creators and aggregators of digital information assets on what elements they should generate or capture as they create or ingest digital information assets. The base profile could thus be used across many digital asset collections at Yale. The base profile, proposed in this report, is a sub-set of the full PREMIS profile that is the main body of this report, and it is intended to be temporary until the library and its peers at Yale have developed digital preservation policies. The Base Profile is presented here. The base profile is intended to apply to objects of all formats. The base profile provides a means of identifying and tracking objects. Small errors and inconsistencies can easily make these functions nontrivial. 3 In order for a repository to scale, basic tracking mechanisms need to work. The base profile may be supplemented incrementally with global updates once Yale-wide preservation policies have been established and communities within the university have developed their own domain-specific profiles. 3 Shirky, C. (2005, December). AIHT: conceptual issues from practical tests. D-Lib Magazine, 11 (2). Retrieved September 29, 2006, from Yale University / Preservation Metadata Task Force Using PREMIS to support preservation of digital assets at Yale / 1-3

4 PREMIS Base Profile The Object Characteristics category records technical properties such as size and format that are applicable to all or most digital formats. In order to insure the fixity of a digital object, i.e. to certify that an object has not been altered in an unauthorized way, two Message Digests, or fixity checks, should be performed automatically by the repository upon ingest and at a later time. The Message Digest Algorithm defines the specific algorithm used. The Message Digest is the output value that can be stored and compared in future fixity checks. 1. objectidentifier objectidentifiertype objectidentifiervalue 2. objectcharacteristics fixity messagedigestalgorithm messagedigest format formatdesignation 2.2.1a. formatname 2.2.1b. formatversion The Format and Format Designation categories accurately and automatically identify file formats upon ingest into a repository. Format Name designates the file format, e.g. Adobe PDF Format Version records the version of the format named in formatname, e.g. 6.0 An Object Identifier should be automatically created by the repository system at the time of ingest. The object identifier serves as the repository's primary identifier in order to insure that identifiers are unique and usable by the repository. Until a repository for long-term preservation is established, creators should assign their own unique identifiers to digital materials intended for longterm preservation. The Object Identifier Type designates the domain within which the object identifier is unique. The Object Identifier Value designates the value of the objectidentifier. Yale University / Preservation Metadata Task Force Using PREMIS to support preservation of digital assets at Yale / 1-4

5 Full profile. This profile would provide guidance to administrators of digital information assets acting as trusted custodians of material deemed to be of long-term value. The full profile is a draft that needs to be fine-tuned through experience with actual instances of use at Yale. We can only go so far based on documents, theories, and the experience others are beginning to have. Experience using PREMIS will determine which elements in the PREMIS model are necessary at Yale. For an illustration of how the full profile may be applied to a particular thing, see the accompanying example that is based on material from the digitization of slide images from the Classics Dept. The full profile is represented in the accompanying spreadsheet that describes the PREMIS model element by element, providing rationale, usage guidelines, and estimates of obligation for each element. Yale University / Preservation Metadata Task Force Using PREMIS to support preservation of digital assets at Yale / 1-5

6 Recommendations Taking action on preservation metadata at Yale will be complex because Yale is a complex organization. The following recommendations call for developing a coherent, sustained effort to preserve digital information assets at Yale by beginning (largely) with the library. Yale can use its library to prototype solutions for policy, technology and processes. The actions for the library named below should be undertaken as the first step in a larger process to successfully guide and support the creation, acquisition, use and preservation of digital information assets at Yale. Experiment with and evaluate PREMIS metadata in use at Yale. o Action: Include PREMIS as part of the Yale University Library VITAL Fedora implementation and to make preliminary report in June 2007 to update IAC on progress. Participate in the PREMIS Implementers Group to work collaboratively with other PREMIS implementers, such as the Library of Congress, the National Library of Australia and Stanford University. o Action: Assign staff involved with the VITAL/Fedora implementation of PREMIS and the library s Digital Preservation Committee to participate in the PREMIS Implementers Group. The implementation group will update IAC in June 2007 on the status of Yale s participation. Establish a coherent policy framework and governance structures for the preservation of digital information assets at Yale. o Action: Digital Preservation Committee (DPC) to develop digital preservation policies for Yale University Library and to make a preliminary report to IAC in June 2007 o Action: Digital Landscape Committee to develop digital preservation policies for Yale University. o Action: IAC and Yale University Library to support use of the base preservation metadata profile throughout Yale by providing critical services to users. Objective: establish persistent naming services for digital assets in Library by June Objective: establish validation services for digital assets in Library by June Objective: establish fixity check services for digital assets in Library by June Yale University / Preservation Metadata Task Force Using PREMIS to support preservation of digital assets at Yale / 1-6

7 Guide to the full profile in the accompanying spreadsheet The PREMIS data dictionary has 22 metadata semantic units or data elements (19 contain nested sub-elements) divided across four types of entities: Object, Event, Agent, and Rights. The body of the evaluation addresses each element and nested sub-element in the PREMIS data dictionary. The full preservation metadata profile is contained in an Excel spreadsheet. All the PREMIS semantic units are broken out as individual elements and sub-units across the four types of PREMIS entities: Object, Event, Agent, Rights. For each element and sub-elements, we suggest a responsible agent, indicate whether the data is likely to be automatically or manually supplied, provide a rationale for the use of the element and, give usage guidelines, and note any applicable data constraints on the value for each element or sub-element. Additionally, we note the applicable object category to which the element can be applied and state the level at which the element is nested within the matrix of PREMIS semantic units. Each of these aspects is explained in more detail below. Semantic Unit: the PREMIS name for the elements used in this evaluation. (PREMIS): the level of obligation assigned in the PREMIS Data Dictionary. Possible values are Mandatory and. (Yale Full Profile): the level of obligation assigned by this group for the full profile. These assignments of obligation are our current best guess and are likely to be adjusted after further experience is gained with use. Possible values are Mandatory,,, and. Responsibility: an indication of who is obliged to provide the metadata. The possible choices are the creator of the digital asset, the repository that is accepting the digital asset, or the submitter. For some elements, more than one responsible party may be designated. Flexibility in designating responsibility is required for at least the following reasons: (1) parties may assume joint responsibility for the generation of certain elements, (2) a repository may assume responsibility for metadata not provided by the creator or submitter if the object is of high value, or (3) designation of responsibility may depend upon relationships among the parties (e.g., if the repository is a client of the creator). Generation: an indication of how the data for this element can be supplied; the only choices indicated here are automated, semi-automated, or manual. Semi-automated processes include use of default settings, batch processes, and templates. Rationale: a brief statement declaring the value of the element to the repository (what functionality the repository gains by knowing this information). Yale University / Preservation Metadata Task Force Using PREMIS to support preservation of digital assets at Yale / 1-7

8 Usage notes: es on best practices for using the element. This may be a rich target for comments as usage may depend greatly on the goals of the administrative unit and the implemented environment of any given repository. Data Constraint: Any constraint on the data, such as integer, controlled vocabulary, container, none. Repeatability: Possible values are repeatable or not repeatable. The value applies to all object categories to which the element is applicable unless stated otherwise. Applicable Object Categories: Indicates which PREMIS model object entities the element can be applied to ( File, and Bitstream). Each PREMIS semantic unit has defined object categories. Some may be applied to all, other to just one of the object categories. Parent: The parent element named is the container closest to the element being described. Yale University / Preservation Metadata Task Force Using PREMIS to support preservation of digital assets at Yale / 1-8

9 Object Entity Semantic Unit (PREMIS) objectidentifier Mandatory Mandatory objectidentifiertype Mandatory Mandatory (Yale Full Profile) Responsibility Generation Rationale Usage es Needed to identify objects and to relate them to other objects, events, and rights. Designates the domain in which the objectidentifier is unique. Data constraint Repeatability Combination of type and value should be globally unique. Container Repeatable Use identifier automatically created by the repository as the primary identifier. Implement a handle system for use as the primary identifier within the repository, and use this value: handle. The handle syntax should include a substring for the collection and sequential suffixes to indicate structural or derivative relationships among objects (e.g., yale.dp/ydc where dp indicates the repository, ydc indicates the collection, and 001 indicates that the object is the original object submitted to the repository). Vocabulary repeatable objectidenifiervalue Mandatory Mandatory Designates the value of the objectidentifier. None repeatable Applicable Object Categories Parent File, Bitstream Object Entity File, Bitstream objectidentifier File, Bitstream objectidentifier preservationlevel Mandatory objectcategory Mandatory Mandatory objectcharacteristics Mandatory Mandatory See components. Use a controlled vocabulary. Assign at the repository or institution level rather than at the object level. Determine according to one or more of the following factors: cultural value, uniqueness, preservability, costs, etc. (e.g., Rutgers Libraries uses two different sets of values: (1) high, medium, low, or none and (2) full, or bitstream). Associate each value for preservationlevel with a profile that designates which semantic units are mandatory and how their values Allows a range of local policies and practices to be should be recorded. Profiles for higher preservation applied differentially to objects or collections within the levels would have more stringent requirements. repository. Requires policy decision by DPC. Vocabulary repeatable For repositories concerned with not only bit-level preservation, but also maintaining usable versions of intellectual entities over time, permits designation of significant properties of one or more sets of digital objects that make up an intellectual entity. Needed to describe such characteristics as See composition level, fixity, size, format, significant components. properties, and inhibitors. Use these values: File, Bitstream. Vocabulary repeatable File Object Entity File, Bitstream Object Entity Used to record technical properties. Format-specific properties are out of scope for PREMIS. Container Repeatable File, Bitstream Object Entity compositionlevel Mandatory Mandatory fixity Mandatory messagedigestalgorithm Mandatory Mandatory Needed to decode or unbundle the object if it is compressed or encrypted. Needed to guarantee object authenticity. Needed to understand the digest value. Supply value even when object is uncompressed and Nonnegative unencrpyted, e.g., assign 0 for base level, 1 for compressed file, 2 for compressed and encrypted file. integer repeatable File, Bitstream objectcharacteristics Investigate cost to generate and maintain message digests calculated by one or more algorithms. The PREMIS Data Dictionary recommends using two or more message digests calculated by different algorithms. Requires policy decision by DPC. Container Repeatable File, Bitstream objectcharacteristics Use registry. Until a global registry becomes available, create and maintain a local registry of encryption algorithms by type: hash algorithms, symmetric algorithms, asymmetric algorithms. Use entries in the hash algorithm registry to populate this semantic unit. Requires policy decision by DPC. Vocabulary repeatable File, Bitstream fixity Yale University / Preservation Metadata Task Force / Using PREMIS to support preservation of digital assets at Yale / 2-1

10 Object Entity Semantic Unit (PREMIS) messagedigest Mandatory Mandatory messagedigestoriginator size Mandatory (Yale Full Profile) Responsibility Generation Rationale Usage es format Mandatory Mandatory See components. formatdesignation Mandatory formatname Mandatory Mandatory fomatversion Mandatory Data constraint Repeatability Applicable Object Categories Required for comparison with value obtained in previous or subsequent fixity checks. None repeatable File, Bitstream fixity Used to validate authenticity of message digests. Record in Object entity or in Event entity. None repeatable File, Bitstream fixity Needed for server administration, planning, etc. formatregistry Mandatory formatregistryname Mandatory Mandatory formatregistrykey Mandatory Mandatory Needed to identify format unambiguously and consistently. Record value obtained from format validation performed during ingest. Use bytes as unit of measurement. Parent Nonnegative integer repeatable File, Bitstream objectcharacteristics Record by value in formatdesignation or by reference in formatregistry. Requires policy decision by DPC. Container repeatable File, Bitstream objectcharacteristics Needed to record name and version of the object format. Use a controlled vocabulary. Container repeatable File, Bitstream format Needed to identify format by name. Needed to specify version of the format type. Vocabulary repeatable File, Bitstream formatdesignation Vocabulary repeatable File, Bitstream formatdesignation Identifies source of format information. Entry for format in registry may provide additional information about the Identify appropriate global registry or develop local format. registry. See Usage es for creatingapplication. Container Repeatable File, Bitstream format Needed to reference the registry by name, URI, etc. Use formal name, local name, or URI consistently. Requires policy decision by DPC. None repeatable File, Bitstream formatregistry Needed to uniquely cite the entry in the register. Requires policy decision by DPC. None repeatable File, Bitstream formatregistry formatregistryrole significantproperties inhibitors Mandatory Indicates type of information provided by registry (e.g., specification, validation), if more than one is used. Requires policy decision by DPC. Used to record properties a repository plans to preserve. See Needed to decrypt files or bypass other protections, components. e.g., passwords. Develop a local controlled vocabulary identifying properties, attributes, or features of objects. Investigate designation of significant properties by profile, e.g., a profile for records might include signatureinformation. None Repeatable Vocabulary repeatable File, Bitstream formatregistry File, Bitstream objectcharacteristics See proposed change < changes.html> Inhibitors may not be detected when a file is parsed. If applicable, require in Submission Information Package (SIP). Container Repeatable File, Bitstream objectcharacteristics inhibitortype Mandatory Mandatory Needed to identify the inhibitor method employed on the object. Use registry. Until a global registry becomes available, create and maintain a local registry of encryption algorithms by type: hash algorithms, symmetric algorithms, asymmetric algorithms. If encryption is used, use entries in the symmetric algorithm subregistry to populate this semantic unit. If password protection is used, use this value: password protection. Requires policy decision by DPC. Vocabulary repeatable File, Bitstream inhibitors Yale University / Preservation Metadata Task Force / Using PREMIS to support preservation of digital assets at Yale / 2-2

11 Object Entity Semantic Unit (PREMIS) inhibitortarget Mandatory inhibitorkey Mandatory (Yale Full Profile) Responsibility Generation Rationale Usage es Needed to identify the content or function is inhibited. Use a controlled vocabulary. Needed to override the inhibitor. Data constraint Repeatability Applicable Object Categories Parent Vocabulary Repeatable File, Bitstream inhibitors Integrate storage of keys with university-wide Public Key Infrastructure (PKI). None repeatable File, Bitstream inhibitors creatingapplication Mandatory creatingapplicationname Mandatory creatingapplicationversion Mandatory datecreatedbyapplication originalname Use registry. Until a global format registry becomes available, create and maintain a local registry of file formats with information, by function (create, render, The creating application may have an effect on identify, validate, etc.), on compatible software. For software and hardware compatibility even if the format each application, provide a standard name, version, is known (cf., Web pages created with Microsoft Front vendor, and system requirements. Use this registry to Page vs. Web pages created with Dreamweaver), which, in turn, is likely to affect migration. populate creatingapplication. Requires policy decision by DPC. Container Repeatable Vocabulary repeatable Compatibility may be specific to versions. None repeatable Used to relate a submitted object and to a derivative object held by the repository. Express dates in the extended format with hyphens. Express times in UTC (Coordinated Universal Time) with the UTC designator ("Z"). ISO 8601 repeatable File, Bitstream Object Entity File, Bitstream creatingapplication File, Bitstream creatingapplication File, Bitstream creatingapplication Use the file name designated in the Submission Information Package (SIP). None repeatable File Object Entity storage contentlocation Mandatory Used to describe the location of two or more bit-wise identical copies of an object, directly under the repository's control, that have the same objectidentifier and are managed as a unit by the repository. Container Repeatable File, Bitstream Object Entity If a handle system is implemented, contentlocation is implicit in objectidentifier. Container repeatable File, Bitstream storage contentlocationtype Mandatory If a handle system is implemented, contentlocationtype is identical to objectidentifiertype. Vocabulary repeatable File, Bitstream contentlocation contentlocationvalue storagemedium Mandatory Mandatory If a handle system is implemented, the information needed to resolve handles to file locations is implicit in the handle system. Documentation for the handle system should be stored by the repository but does not need to be recorded in the metadata for each object. None repeatable File, Bitstream contentlocation Implicit. Determined by repository architecture. Vocabulary repeatable File storage environment Mandatory See components. environmentcharacteristic Used to describe dependent files and hardware/software combinations required to render or See interact with content. Applies to Representation and File, i.e., if the repository has made a commitment to preserve usable versions of intellectual entities over time. May be omitted if the repository is only Record information on functions supported by hardware and software in a registry, not in the metadata for each object. See Usage es for creatingapplication. Record information on dependent components. responsible for bit-level preservation. files in the metadata for each object. Container Repeatable Used to distinguish levels of functionality provided by hardware/software combinations if more than one combination supports a particular purpose. Vocabulary repeatable File, Bitstream Object Entity File, Bitstream environment Yale University / Preservation Metadata Task Force / Using PREMIS to support preservation of digital assets at Yale / 2-3

12 Object Entity Semantic Unit environmentpurpose environmente (PREMIS) (Yale Full Profile) Responsibility Generation Rationale Usage es dependency Mandatory dependencyname Mandatory dependencyidentifier Mandatory dependencyidentifiertype Mandatory Mandatory dependencyidentifiervalue Mandatory Mandatory software swname swversion swtype swotherinformation swdependency Mandatory Mandatory Manual Used to designate how an object may be manipulated using a particular combination of software and hardware. Data constraint Repeatability Vocabulary Repeatable Used to provide information on the environment that cannot be recorded elsewhere. None Repeatable Needed for complex objects that depend on other objects. Use for non-executable components of an object, e.g., a font, style sheet, or schema. Do not use for software or hardware. Container Repeatable Clarifies nature of dependency if not made apparent by the dependencyidentifier. None Repeatable Identifies a dependent object. Use objectidentifier of dependent object. Container Repeatable Designates the domain in which the dependencyidentifier is unique. Vocabulary repeatable Designates the value of the dependencyidentifier. None repeatable The software needed to render the object Software can be inferred from creatingapplication and does not need to recorded in the metadata for each object. See Usage es for creatingapplication. Container Repeatable Needed to identify the manufacturer and name of software product. None repeatable Needed to identify the version number of the software product. Needed to identify the role of the software product e.g., operating system, renderer, driver, etc. None Repeatable (File); Repeatable ( Bitstream) Vocabulary repeatable Used to describe additional requirements or instructions related to the software product. None Repeatable Used to identify and additonal software needed by that in swname. None Repeatable Applicable Object Categories Parent File, Bitstream environment File, Bitstream environment File, Bitstream environment File, Bitstream dependency File, Bitstream dependency File, Bitstream dependencyidentifier File, Bitstream dependencyidentifier File, Bitstream environment File, Bitstream software File, Bitstream software File, Bitstream software File, Bitstream software File, Bitstream software hardware The hardware needed to render or use the object Hardware can be inferred from creatingapplication and does not need to be recorded in the metadata for each object. See Usage es for creatingapplication. Container Repeatable File, Bitstream environment hwname Mandatory Needed to identify the manufacturer, model and version of hardware product. None repeatable File, Bitstream hardware hwtype Mandatory Needed to identify the class or category of the hardware. Vocabulary repeatable File, Bitstream hardware hwotherinformation Used to describe additional requirements or instructions related to the hardware product. None Repeatable File, Bitstream hardware Yale University / Preservation Metadata Task Force / Using PREMIS to support preservation of digital assets at Yale / 2-4

13 Object Entity Semantic Unit (PREMIS) signatureinformation signatureinformationencoding Mandatory Mandatory signer (Yale Full Profile) Responsibility Generation Rationale Usage es See Used to confirm the origin and integrity of objects components. signed by the repository. Used to record encoding of signaturevalue and keyinformation. Needed to determine who created the signature. Data constraint Repeatability Applicable Object Categories Parent Do not use to record information about digital signatures that authenticate agents; use the Event Entity instead. Container Repeatable File, Bitstream Object Entity Use registry. Vocabulary repeatable File, Bitstream signatureinformation Use the name provided in the Submission Information Package (SIP). None repeatable File, Bitstream signatureinformation Use registry. Create and maintain a local registry of encryption algorithms by type: hash algorithms, symmetric algorithms, asymmetric algorithms. Use entries in the asymmetric algorithm registry and hash algorithm registry to populate signaturemethod. Record the encryption algorithm for the signature first (the asymmetric algorithm), followed by a hyphen, followed by the hash algorithm, e.g., DSA-SHA1. signaturemethod Mandatory Mandatory Needed for signature encryption and validation. Requires policy decision by DPC. Vocabulary repeatable File, Bitstream signatureinformation signaturevalue Mandatory Mandatory The value of the digital signature None repeatable File, Bitstream signatureinformation signaturevalidationrules Mandatory Mandatory Used to maintain documentation on signature algorithms, canonicalization algorithms, transform algorithms, etc. signatureproperties Used to verify the authenticity of the signature. keyinformation keytype Mandatory Mandatory keyvalue Mandatory Mandatory Used to record information about the signer's public key, for comparison with the and verifiation of the digital signature. Used to record the encryption algorithm or certification method for authenticating the signer's public key. keyverificationinformation Additional information used to verify the signer's key relationship Mandatory relationshiptype Mandatory Mandatory relationshipsubtype Mandatory Mandatory May be a pointer to external documentation. See < Requires policy decision by DPC. None repeatable File, Bitstream signatureinformation Define suitably granular structure for time of signature generation, serial number of cryptographic hardware used, etc., as needed. Requires policy decision by DPC. None Repeatable File, Bitstream signatureinformation See proposed change at < Container repeatable File, Bitstream signatureinformation Use registry. Requires policy decision by DPC. Vocabulary repeatable File, Bitstream keyinformation The value of the signer's public key is needed for comparison with the signature. None repeatable File, Bitstream keyinformation May be used to record information about certificates, etc. If possible, use keytype instead. Requires policy decision by DPC. None repeatable File, Bitstream keyinformation See components. Needed to represent relationships between objects. Record all relevant relationships. Container Repeatable Structural or derivational relationship between objects Develop a local controlled vocabulary of relationships. Describes the role of each object in the relationship. Include in a local controlled vocabulary of relationships. Vocabulary repeatable Vocabulary repeatable relatedobjectidentification Mandatory Mandatory Identifies related objects. Use objectidentifier of related resource. Container Repeatable relatedobjectidentificationtype Mandatory Mandatory Designates the domain in which the relatedobjectidentification is unique. Vocabulary repeatable File, Bitstream Object Entity File, Bitstream relationship File, Bitstream relationship File, Bitstream relationship File, Bitstream relatedobjectidentification Yale University / Preservation Metadata Task Force / Using PREMIS to support preservation of digital assets at Yale / 2-5

14 Object Entity Semantic Unit (PREMIS) (Yale Full Profile) Responsibility Generation Rationale Usage es Data constraint Repeatability relatedobjectidentifiervalue Mandatory Mandatory Designates the value of the relatedobjectidentification. None repeatable relatedobjectsequence Mandatory Mandatory Needed to know the order of components or files. relatedeventidentification Mandatory Mandatory relatedeventidentifiertype Mandatory Mandatory If there is only one related object, assign the value 0 (zero). If objects are unordered, assign each the same value. None repeatable Used to identify an event that relates the object to a second object, e.g., migration. Use eventidentifier of related event. Container Repeatable Designates the domain in which the relatedeventidentification is unique. relatedeventidentifiervalue Mandatory Mandatory Designates the value of the relatedeventidentifier relatedeventsequence Mandatory Needed to know the order of events. linkingeventidentifier Mandatory linkingeventidentifiertype Mandatory Mandatory Vocabulary repeatable Vocabulary repeatable If there is only one related event, assign the value 0 (zero). If events are unordered, assign each the same value. None repeatable Used to identify an event that does not result in the creation of a second object, e.g., validation. Use eventidentifier of linking event. Container Repeatable Designates the domain in which the linkingeventidentifier is unique. linkingeventidentifiervalue Mandatory Mandatory Designates the value of the linkingeventidentifier. linkingintellectualentityidentifier linkingintellectualentitytype Mandatory Mandatory linkingintellectualentityvalue Mandatory Mandatory Vocabulary repeatable Vocabulary repeatable Use to identify an object whose content is related to the object designated by the objectidentifier. Used to link an object to another object or collection ly, use to identify a metadata record for the that represents the same intellectual entity or a related object designated by the objectidentifier. Requires intellectual entity. policy decision by DPC. Container Repeatable Designates the domain in which the linkingintellectualentityidentifier is unique. Vocabulary repeatable Designates the value of the linkingintellectualentityidentifier. None repeatable linkingpermissionstatementidentifier Used to link an object to a permission statement. Use permissionstatementidentifier. Container Repeatable linkingpermissionstatementidentifiertype Mandatory Mandatory linkingpermissionstatementidentifiervalue Mandatory Mandatory Designates the domain in which the linkingpermissionstatementidentifier is unique. Vocabulary repeatable Designates the value of the linkingpermissionstatementidentifer. None repeatable Applicable Object Categories Parent File, Bitstream relatedobjectidentification File, Bitstream relatedobjectidentification File, Bitstream relationship File, Bitstream relatedeventidentification File, Bitstream relatedeventidentification File, Bitstream relatedeventidentification File, Bitstream Object Entity File, Bitstream linkingeventidentifier File, Bitstream linkingeventidentifier File, Bitstream Object Entity File, Bitstream linkingintellectualentityidentifier File, Bitstream linkingintellectualentityidentifier File, Bitstream Object Entity File, Bitstream linkingpermissionstatementidentifier File, Bitstream linkingpermissionstatementidentifier Yale University / Preservation Metadata Task Force / Using PREMIS to support preservation of digital assets at Yale / 2-6

15 Event Entity Semantic Unit (PREMIS) eventidentifier Mandatory Mandatory (Yale Full Profile) Responsibility Generation Rationale Usage es Data constraint Repeatability Parent Needed to identify events and to relate them to objects and agents. None repeatable Event Entity Use identifiers automatically created by the repository. If information needs to be recorded on events outside the repository, use a registry of repositories. (Each eventidentifiertype Mandatory Mandatory Designates the domain in which the eventidentifier is unique. repository is likely to implement its own numbering system for events.) None repeatable eventidentifier eventidentifiervalue Mandatory Mandatory Designates the value of the eventidentifier. None repeatable eventidentifier eventtype Mandatory Mandatory Semi-automated Used to categorize events that affect the preservation of digital objects. limited to preservation treatments. eventdatetime Mandatory Mandatory Required for digital provenance. Manual Develop local controlled vocabulary of event types. Requires policy decision by DPC. Vocabulary repeatable Event Entity Express dates in the extended format with hyphens. Express times in UTC (Coordinated Universal Time) with the UTC designator ("Z"). ISO 8601 repeatable Event Entity Additional information about an event; not intended to be machine processible None repeatable Event Entity eventdetail Machine-processible information about event eventoutcomeinformation Mandatory See components. outcomes None Repeatable Event Entity Develop local controlled vocabulary of event eventoutcome Mandatory Coded representation of the outcome of an event outcomes. Requires policy decision by DPC. Vocabulary repeatable eventoutcomeinformation Detailed description of the outcome of an event; may Establish guidelines for what to record for each event eventoutcomedetail Semi-automated be used to record or to point to event logs type. Requires policy decision by DPC. None repeatable eventoutcomeinformation linkingagentidentifer Mandatory Required for digital provenance. Use agentidentifier. None Repeatable Event Entity Designates the domain in which the agentidentifier is linkingagentidentifertype Mandatory Mandatory unique. Vocabulary repeatable linkingagentidentifier linkingagentidentifiervalue Mandatory Mandatory Designates the value of the agentidentifier. None repeatable linkingagentidentifier Used to disambiguate agent role if more than one agent is associated with an event or if the role of the linkingagentrole agent is not evident from the eventtype. Vocabulary Repeatable linkingagentidentifier linkingobjectidentifier Mandatory Required for digital provenance. Use objectidentifier. None Repeatable Event Entity Designates the domain in which the objectidentifier is linkingobjectidentifiertype Mandatory Mandatory unique. Vocabulary repeatable linkingobjectidentifier linkingobjectidentifiervalue Mandatory Mandatory Designates the value of the objectidentifier. None repeatable linkingobjectidentifier Yale University / Preservation Metadata Task Force / Using PREMIS to support preservation of digital assets at Yale / 2-7

16 Agent Entity Semantic Unit (PREMIS) (Yale Full Profile) Responsibility Generation Rationale Usage es Data constraint Repeatability Parent Needed to identify agents, to relate them to events, and to designate their permissions. Create a local registry; create a registry for external agents as needed. For the local registry, use Yale University netids and integrate with CAS. None Repeatable Agent Entity agentidentifier Mandatory Mandatory Designates the domain in which the agentidentifier is agentidentifiertype Mandatory Mandatory unique. Vocabulary repeatable agentidentifier agentidentifiervalue Mandatory Mandatory Designates the value of the agentidentifier. None repeatable agentidentifier agentname agenttype The agent may not be known by his/her agentidentifier. Record the agent's name in the event that the agent's CAS account does not persist over the lifetime of objects related to events in which the agent played a role. None Repeatable Agent Entity High-level characterization of the type of agent, useful for filtering, etc. Develop a local controlled vocabulary of agent types. Vocabulary repeatable Agent Entity Yale University / Preservation Metadata Task Force / Using PREMIS to support preservation of digital assets at Yale / 2-8

17 Rights Entity Semantic Unit (PREMIS) (Yale Full Profile) Responsibility Generation Rationale Usage es Data constraint Repeatability Parent permissionstatement Mandatory permissionstatementidentifier Mandatory Mandatory Contains all other semantic units in the Rights Entity. Used to designate the rights of agents with respect to objects held by the repository. If a permissionstatement is not provided, then a linkingpermissionstatementidentifier is not provided in the Object Entity. Investigate objects or collections, if any, for which a permissionstatement is required and those for which permissions may be set by institutionwide policy. Container Repeatable Rights Entity Needed to provide unique identifer for permission statement so that it can be related to obects, agents and events. Container repeatable permissionstatement permissionstatementidentifiertype Mandatory Mandatory Designates domain in which the permissionstatementidentifier is unique. Consider rights management standards informed by the Digital Millennium Copyright Act. Also, see JISC. Digital Rights Management: Final Report < Vocabulary repeatable permissionstatementidentifier permissionstatementidentifiervalue Mandatory Mandatory Designates value of the permissionstatementidentifier. None repeatable permissionstatementidentifier Investigate the appropriate level for designating linkingobject Mandatory Mandatory Identifies the object(s) related to the permission statement. permissions, e.g., by object, file type, agent, or collection. None Repeatable permissionstatement grantingagent Identifies the agent granting permission. Use agentidentifier. None Repeatable permissionstatement grantingagreement, Refers to a document or contract recording the granting of permission. grantingagreementidentification Identifies the agreement that grants permissions. For objects or collections for which a permissionstatement is required, include a grantingagreement in the Submission Information Package (SIP). None repeatable permissionstatement If a permissionstatement is required, ingest the Submission Information Package (SIP) and use the objectidentifier of the SIP. None repeatable grantingagreement grantingagreementinformation, Text describing the agreement If a permissionstatement is required, optionally paraphrase or quote the text of the grantingagreement provided in the Submission Information Package (SIP). None repeatable grantingagreement permissiongranted Mandatory Mandatory Identifies the allowed actions per permission. None Repeatable permissionstatement act Mandatory Mandatory Details the actions that are allowed. Develop a local controlled vocabulary of acceptable acts. ly, define a single controlled vocabulary for eventtype and act. Vocabulary repeatable permissiongranted Identifies the limits on actions (e.g., the number of Include any restrictions in a local controlled vocabulary restriction copies that can be made). of acceptable acts. None Repeatable permissiongranted termofgrant Mandatory Mandatory Declares the term of the grant. None repeatable permissiongranted startdate Mandatory Mandatory Needed to know the start of term. Record if term of permissions is time bounded. The datatype of this semantic unit is xs:date, which has year, month, and day properties, plus an optional timezone-valued timezone property. Express as a nontimezoned value in the extended format: YYYY- MM-DD. Year, month, and day are required. ISO 8601 repeatable termofgrant Yale University / Preservation Metadata Task Force / Using PREMIS to support preservation of digital assets at Yale / 2-9

18 Rights Entity enddate Mandatory Mandatory Needed to know end of term. Record if term of permissions is time bounded. The datatype of this semantic unit is xs:date, which has year, month, and day properties, plus an optional timezone-valued timezone property. Express as a nontimezoned value in the extended format: YYYY- MM-DD. Year, month, and day are required. ISO 8601 repeatable termofgrant permissione Provides additional information on permissions. Use only if information cannot be recorded elsewhere. None Repeatable permissiongranted Yale University / Preservation Metadata Task Force / Using PREMIS to support preservation of digital assets at Yale / 2-10