A | B | C | D | E | F | G | H | I | J | K | L | M | N | O | P | Q | R | S | T | U | V | W | X | Y | |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
1 | Stanford priority | Cornell priority | Harvard priority | Timing / Scope | Requirement | Category | Requirement comes from | Questions and comments | Stanford comments | Cornell comments | Harvard comments | ||||||||||||||
2 | -- | highest | high | 2017Q1 | Take MARCXML file for one bib record as input, output LD4L Ontology RDF | A. core | Cornell UC1 | For LD4P cataloging/extension work. No expectation of deduping across a collection of records, or of reconciliation. Architecture to assume later extension to different input and outputs. Maps to Harvard UC1. | Stanford requires conversion to Bibframe 2.0, not LD4L ontology. | ||||||||||||||||
3 | high | medium | medium | 2017Q2? | Deduplicate URIs generated in conversion of a batch of records | A. core | Stanford, Cornell UC5, Harvard UC5 | Understanding of context? Batch or something else? This is a high priority (Cornell) but recognize that it happens after the "high" priorities listed in column B | |||||||||||||||||
4 | high | high | high | 2017Q1 | Convert from MARC (as MARCXML) | B. source data (input) | Stanford, Cornell UC1 | ||||||||||||||||||
5 | low | low | post LD4L Labs | convert from MODS | B. source data (input) | Stanford | convert SDR legacy data? SDR metadata originally created not in MODS--does Stanford need this? | ||||||||||||||||||
6 | medium | medium-to-low | medium? | convert source data provided via post to web service (so that the converter can be a function of an editor) | B. source data (input) | converter discussion at Stanford LD4P core mtg | SF- depends expected workflows (whether items are edited on their way into the knowlege store, or after). Would like to hear Stanford's thoughts on this. JAK: for Cornell, seems essential moving forward for production environment but not needed for current LD4P effort | ||||||||||||||||||
7 | low | low | medium | convert directly from database (converter reads database) | B. source data (input) | converter discussion at Stanford LD4P core mtg; Harvard UC2 | is this a functional requirement? is there a reason for this beyond "better scaling and efficiency" (a good reason, but not a functional requirement). SF- may not be a functional requirement for every workflow, but there will be workflows where we can't devote resources to exporting files from a db, to be converted periodically. Hoping to explore this with the Harvard Film Archive db. | ||||||||||||||||||
8 | high | high | high | 2017Q2? | convert a batch of source data (up to millions of records) | B. source data (input) | Stanford Conversion Workflow Use Cases – MARC, Cornell UC4, Harvard UC4 and UC5 | Cornell UC4 imagines doing this as input and output files (at first at least) | dependency for original cataloging | how is this different from the first functional requirement around converting a batch of MARCXML? Is the idea that this is non-MARC data? OR that idea of scalability to the millions? This is not required for the LD4P Cornell project but is the core of the functionality of the converter so high | |||||||||||||||
9 | high | high (ld4l BF ext); low (BF) | 2017Q1 | convert a single record (especially to use as the basis for original cataloging) | B. source data (input) | Stanford Conversion Workflow Use Cases – MARC | how does this relate to 7D? | Same as line 2 but for BF2 | |||||||||||||||||
10 | -- | high | 2017Q1 | convert even brief records such as initial record from acquisitions | B. source data (input) | tracer_bullet_workflow_user_stories | from converter point of view, how is this different from converting a "full record", what does the converter have to do differently? | Suggest to remove this requirement | how is this a different requirement than convert a full record? If anything, this is facilitated in the former. | ||||||||||||||||
11 | -- | low | high | 2017Q2 | Take one FGDC file as input, output LD4L Ontology RDF | B. source data (input) | Cornell UC2 | For LD4P cataloging/extension work. Expect Cornell-Harvard collaboration. Harvard UC3 | While FGDC is listed as an example, this is a Harvard requirement rather than a Cornell requirement in actual practice | This is on Harvard's roadmap to work on during 2017Q2. This is dependent on core converter handling all common marc/fgdc fields by 4/1. | |||||||||||||||
12 | -- | low | high | 2017Q3 | Take one HFA file as input, output LD4L Ontology RDF | B. source data (input) | Cornell UC2 | For LD4P cataloging/extension work. Expect Cornell-Harvard collaboration. Harvard UC2 | I'm unsure why HFA data is a Cornell priority -- we do not have data for the Harvard Film Archive | This is on Harvard's roadmap to work on during 2017Q2. | |||||||||||||||
13 | high | low | high | when a single MARC record represents both the print and digital version of a resource, converter must correctly interpret whether there are 2 works and 1 instance of each or 2 instances of the same work | B. source data (input) | Stanford | is this part of converter mapping? SF- From Harvard's POV we will have to look at our single record practice and possibly create a mapping for our local needs. | See local tag use cases tab for 856 | Cornell generally separates physical and digital bibs aside from known areas (e.g.: dissertations) | ||||||||||||||||
14 | -- | high | high | 2017Q1 | Convert to LD4L-Ontology | C. target data (output) | Stanford LD4P Converter Requirements3, Cornell UC1 | ||||||||||||||||||
15 | high | medium-to-low | medium | 2017Q2? | Convert to BF 2.0 | C. target data (output) | Stanford, https://wiki.duraspace.org/display/ld4lPLAN/Cornell+Converter+Use+Cases#CornellConverterUseCases-UC3 | At 2016-11 Cornell meeting was agreed to do LD4L Ontology first, expect to be able to readily map from internal representation to BF2 as a change in the output code | |||||||||||||||||
16 | medium | Output triples to include inverses for where both subject and object are in target domain | C. target data (output) | Discussion in 2017-02-07 converter meeting | Could also be done as a post-process step | ||||||||||||||||||||
17 | low | high-to-medium | high | convert to a core ontology with an extension | C. target data (output) | Stanford | Not a tracer-bullet 1 (coversion) requirement | ||||||||||||||||||
18 | low | low | converter output, you should be able to choose your serialization format, or, hook converter to a serializer in which case, what format do you want from converter? do you care? | C. target data (output) | Stanford | SW - Conversion between RDF serialization formats is relatively trivial and supported by many off-the-shelf tools. Might be best as a post-process step although since Java libraries also support different formats it would likely be easy to build in some selection | There are other tools to be able to reserialize RDF downstream if needed | ||||||||||||||||||
19 | low | low | The converter needs to be configurable to put or post the RDF somewhere (rather than to standard out) | C. target data (output) | Stanford | We can deal with stdout | |||||||||||||||||||
20 | medium | medium | high | be able to call converter in various ways (API?) at various pipeline steps, including from an editor-like tool in order to review converter output; call converter from another process | D. pipeline | converter discussion at Stanford LD4P core mtg | Do you want the converter to be able to launch another process? SW - Could be implemented as a wrapper around this or another converter | Related to line 7 | |||||||||||||||||
21 | low | low | put the output into a triple store | D. pipeline | Stanford Conversion Workflow Use Cases – MARC | Related to 20 | Is this in scope for the converter per-se, or is this something one does with converter output? | ||||||||||||||||||
22 | ? | low | Reconcile with external identities | E. reconciliation | Stanford, Cornell UC5 | Automated only, manual assist out-of-scope. Stanford question whether this encompases all entity lookup | Not sure what this mean exactly. Is this a post-process or is this a converter process via "uri minting" | recon was a separate step according to our understanding | |||||||||||||||||
23 | low | low | medium | convert newly received data relevant to previously converted records (MARC record is updated post-conversion) | G. updates | Stanford, Harvard UC5 | ideally the converter would convert against something and tell you if it found conflicts and let you choose what to do; converter would ignore dupes, not put them in triple store. if this is a situation that is going to be temporary, need to build converter for it? | Better handled downstream | |||||||||||||||||
24 | low | low | medium | Take daily dumps of changed MARCXML records, convert these to LD4L ontology and update a triplestore of the complete catalog data | G. updates | Cornell UC6 | Also has performance implications, Harvard UC6 | ||||||||||||||||||
25 | low | medium | Say what conversion process (profile) was applied | H. provenance tracking | Stanford | ||||||||||||||||||||
26 | low | low | Say where the data came from | H. provenance tracking | Stanford | Before we acquired data? Or from which catalog the source data for the converter derived? | |||||||||||||||||||
27 | high | high | high | 2017Q2? | If a record can't be converted, converter skips the record and continues (and reports the skipped records and reasons for skipping in a convenient format) | I. error handling | Stanford LD4P Converter Requirements3, Cornel UC4 | Based on experience with converter used in LD4L1 | |||||||||||||||||
28 | high | high | high | 2017Q2? | Report errors with enough detail that someone can fix the problem records (which records were skipped and which fields caused the problem) | J. reporting | Stanford, Cornell UC4 | ||||||||||||||||||
29 | low | medium | on completion of conversion get performance statistics including: # of records converted; # of records skipped; time elapsed; # of local URIs created; # of dedpues; # of entities (what?); # of triples created; # of errors, what kind of errors; what kind of reconciliation was done, to what sources, how many things were reconciled, how many were queued for human intervention; what didn't convert (at record level or field level?) | J. reporting | Stanford | ||||||||||||||||||||
30 | low | low | at conversion time, specify where the converted data goes (including putting it in a temporary location for review) | K. configuration | Stanford Conversion Workflow Use Cases – MARC | ||||||||||||||||||||
31 | medium | medium | at conversion time, choose a target ontology and optionally, target ontology extension | K. configuration | Stanford Conversion Workflow Use Cases – MARC | ? Question about ld4l base conversion ontology and its ability to subsequently map to BF2 and/or other ontology extentions - jg | |||||||||||||||||||
32 | high | high | high | 2017Q2? | Specify how converter will handle local MARC practice | K. configuration | Stanford Conversion Workflow Use Cases – MARC | are these tags that no one else uses, or tags that are used in a particular way locally, or a local preference for how to convert tags that are used in a standard way? SF- likely both; seems to be about the flexibility/extensiblity of the converter | |||||||||||||||||
33 | high | high-to-medium | create profile, which includes...how to handle private data (and which data is private), how to handle local fields, what target ontology/extension to use, where to put output, ... | K. configuration | Stanford LD4P Converter Requirements3 | ||||||||||||||||||||
34 | high | medium | medium | save profile | K. configuration | Stanford LD4P Converter Requirements3 | |||||||||||||||||||
35 | low | medium | medium | share profile | K. configuration | Stanford LD4P Converter Requirements3 | |||||||||||||||||||
36 | low | medium | medium | another part of a profile is to specify some constant values to be added to all records being converted (some triple statements to create for everything?) examples ... | K. configuration | Stanford | SF- one use case we have for this is to assert an LDN Receiver for specific types of things. | ||||||||||||||||||
37 | high | high | high | 2017Q1 | URIs are minted in institution's local namespace | K. configuration | Stanford | Expectation that converter running at any institution will be configured with the local namespace to use for URI minting | |||||||||||||||||
38 | high | medium | high | add new ontologies or ontology extensions later as conversion targets | L. extensibility | Stanford | yes it is a priority (high) but not urgent to facilitate first stage of conversion dev aside from knowing the converter needs to be extensible | ||||||||||||||||||
39 | low | -- | performance specs: convert 10 million records in less than one week | M. non-functional | converter discussion at Stanford LD4P core mtg | ||||||||||||||||||||
40 | low | -- | run multiple conversions in parallel (i.e. converter is not tied up with one job) | M. non-functional | Stanford | ||||||||||||||||||||
41 | high | high | ongoing | Documentation for users and for developers | M. non-functional | Stanford LD4P Converter Requirements3 | SW - Documentation should not be a separate task, should document as features are added. Might need a specific effort around end of grant to to tidy though. | ||||||||||||||||||
42 | low | low-to-no | post LD4L Labs | nice to have: converter chooses ontology and extension for you (on record by record basis?) based on source data (and a set of configurable rules) | N. bluesky | converter discussion at Stanford LD4P core mtg | |||||||||||||||||||
43 | low | low | what happens to previously converted records if/when the ontology you are using gets updated? | N. bluesky | tracer_bullet_workflow_user_stories | Ontology change management (cf. Javed's work), probably out-of-scope for this grant | Seems like less of a converter issue and more of a long-term operations workflow, procedure and infrastructure question | ||||||||||||||||||
44 | low | low | GUI to set configurations (anticipating the need for non-programmer interactions) | N. bluesky | Harvard | ||||||||||||||||||||
45 | low | -- | Converter recognition of uploaded target ontology/extensions | N. bluesky | Harvard | SF-related to "add new ontologies or ontology extensions later as conversion targets", but wasn't sure if that meant adding code that specified the ontology classes or properties or having the ontology "loaded" to the converter and referenced in the profile. | Confused why asking about ontology (in converter context) rather than mapping between source data and output ontology... does ontology need to be loaded? | ||||||||||||||||||
46 | |||||||||||||||||||||||||
47 | |||||||||||||||||||||||||
48 | |||||||||||||||||||||||||
49 | |||||||||||||||||||||||||
50 | |||||||||||||||||||||||||
51 | |||||||||||||||||||||||||
52 | |||||||||||||||||||||||||
53 | |||||||||||||||||||||||||
54 | |||||||||||||||||||||||||
55 | |||||||||||||||||||||||||
56 | |||||||||||||||||||||||||
57 | |||||||||||||||||||||||||
58 | |||||||||||||||||||||||||
59 | |||||||||||||||||||||||||
60 | |||||||||||||||||||||||||
61 | |||||||||||||||||||||||||
62 | |||||||||||||||||||||||||
63 | |||||||||||||||||||||||||
64 | |||||||||||||||||||||||||
65 | |||||||||||||||||||||||||
66 | |||||||||||||||||||||||||
67 | |||||||||||||||||||||||||
68 | |||||||||||||||||||||||||
69 | |||||||||||||||||||||||||
70 | |||||||||||||||||||||||||
71 | |||||||||||||||||||||||||
72 | |||||||||||||||||||||||||
73 | |||||||||||||||||||||||||
74 | |||||||||||||||||||||||||
75 | |||||||||||||||||||||||||
76 | |||||||||||||||||||||||||
77 | |||||||||||||||||||||||||
78 | |||||||||||||||||||||||||
79 | |||||||||||||||||||||||||
80 | |||||||||||||||||||||||||
81 | |||||||||||||||||||||||||
82 | |||||||||||||||||||||||||
83 | |||||||||||||||||||||||||
84 | |||||||||||||||||||||||||
85 | |||||||||||||||||||||||||
86 | |||||||||||||||||||||||||
87 | |||||||||||||||||||||||||
88 | |||||||||||||||||||||||||
89 | |||||||||||||||||||||||||
90 | |||||||||||||||||||||||||
91 | |||||||||||||||||||||||||
92 | |||||||||||||||||||||||||
93 | |||||||||||||||||||||||||
94 | |||||||||||||||||||||||||
95 | |||||||||||||||||||||||||
96 | |||||||||||||||||||||||||
97 | |||||||||||||||||||||||||
98 | |||||||||||||||||||||||||
99 | |||||||||||||||||||||||||
100 |