Changes between Version 3 and Version 4 of Taskforces/FCS/VidConf20150624


Ignore:
Timestamp:
06/24/15 12:56:17 (9 years ago)
Author:
Leif-Jöran
Comment:

draft minutes

Legend:

Unmodified
Added
Removed
Modified
  • Taskforces/FCS/VidConf20150624

    v3 v4  
    1313 1. !DataView(s)
    1414
    15 == Outcome: Minutes, Decisions and Actions ==
    16  will be here today.
     15== Outcome: Minutes, Decisions and Actions (currently draft) ==
     16 Main focus Dataview return format
     17
     18Not much examples except for Oliver's.
     19
     20Advanced dataview
     21Adhoc or based on something?
     22Oliver mental model one dataview will convey all layers etc
     23
     24So no msd dataview
     25explict elements for some layers
     26container format.
     27Pavel, looks nice. Reason behind it?
     28Any tree? Complex tree view? Multiple layers?
     29
     30Press stop 2-3 secs after stop talking.
     31Hierarchical structure.
     32Proper scope of work.
     33Transport hierarchy
     34Matej, really another complexity level. Start with query annotation layers, and only when we master that move forward.
     35
     36Pavel, second suggestion multi annotations. Fine with forgetting trees for now. structured/hierarchical attributes.
     37I think it can be nice in terms of ease of understanding that some thing belong together.
     38
     39Matej, what do you mean? Example.
     40
     41case
     42
     43Oliver (IDS): question of course is, if structured, how would it be structured ;)
     44a client like the aggregator needs to make sense of that
     45
     46harmonize if they are structured.
     47No good solution yet.
     48
     49Dieter: Pavel: is there a standard way for that in the universal dependencies?
     50
     51Controlled vocab
     52Like for POS the UD-17 we decided on.
     53Jörg not every endpoint will provide UD-17 we are not going to retag all corpora.
     54
     55Pavel, agree hierarchical attributes is too complex for now.
     56
     57Jörg bidrectional translation.
     58
     59Dieter, limited list
     60choice
     61first proposal, syllables.
     62finer granularity. In worst case every single character.
     63
     64Candidate for speech
     65or primarily for textual resources?
     66
     67time lined signal
     68"textual" annotations
     69
     70Dieter concrete example. Silence is best covered by example 1.
     71Transcribed speech corpora.
     72
     73Matej, generic way to describe atomic units. So time stamp could be fitted with example 2.
     74
     75Oliver, yes, offsets can well be timestamps
     76
     77silence background noice
     78
     79Combination of dataviews
     80allow to reference other layer items
     81non-superior layers
     82order is not defined, but
     83Dieter: I have the impression we are re-inventing formats like EAF: https://corpus1.mpi.nl/media-archive/demo/Ams_Demo/versioning_demo/Annotations/118_fishing2-fire-2011.eaf
     84
     85Comparison matrix.
     86
     87Keep things simple.
     88
     89Keep it fairly straight forward.
     90
     91Keep non-textual formats in mind.
     92
     93Doing concrete proposals. More are needed.
     94
     95transform to standard format.
     96
     97What about parallel corpora?
     98another annotation layer trans
     99
     100More discussion is needed. Concrete proposals.
     101
     102Hanna, concrete examples? Propose new layers?
     103Mappings are welcome.
     104
     105Use second view in examples
     106
     107Need attribute for unit.
     108
     109Matej, lexical resources.
     110
     111Toying
     112
     113
     114LJO: add page for advanced datawiew to trac.
     115Next meeting, 2-3 weeks
     116Might be flash or adobe connect
     117
    17118
    18119== Documents ==