VLO Taskforce Feedback of the F-AGs

Notes:

• F-AGs are domain specific working groups supporting CLARIN-D.
• To view the original German version of this page use the page history.

F-AG 6 (Petra Wagner)

(Regarding Facet 11: Modality)

»I might be missing something obvious, but I consider mixing information on the communication channel (for the signal: audio, video, text) and information on the modality (spoken vs. written language, gesture) as problematic. The channel is essentially completely agnostic with respect to the modality that is transmitted. I. e. the visual channel transmits articulatory, hand, body or head gestures, maybe also dance, images, texts or even the process of text production (writing). Gesture can be transmitted by different channels, e. g. video, EMA, MoCap?. I think we'll need more details to fully understand the proposed vocabulary. I'd like to further discuss this issue with the VLO-TF.

F-AG 10 (Thomas Werneke)

• facets should be visible on the start page
• user documentation should be available from the start page (or at least a link should be provided); a short introductory note would be nice
• search results:
  • should be always sorted/ranked/marked, e .g.:
    • using a dedicated color scheme
    • by clustering according to a specific set of criteria
    • by providing an icon representing the resource type for each result
  • users should see more than 10 results
  • in the result list, the entries should be more self describing; resource type, genre, domain should be visible on first sight
  • hierarchical ranking of the results is important; ordering by frequency can only be seen as a first quick and dirty approach
• a facet for time coverage will be important and useful
• extended search interface would be useful, e. g. allowing queries for all newspapers
• a possibility to create your own collections would be nice

Feedback from F-AG leaders, meeting 15th May 2015

• trade-off between users' domain specific search demands and the general purpose presentation of resources in the VLO → search facilities should be more narrowly targeted to different scientific domains
• current sorting leads to result lists that are difficult to understand and use, e. g. as returned by a search for »Germanistik« as keyword
• clustering of results an option? e. g. grouping of single resources that are part of bigger collections (»Wortschatz Leipzig«)
• some resources appear twice due to the LINDAT-CZ mirror
• How can I confirm with certainty whether a resource is already in the VLO or not? (note from the translator: This question probably aims at duplicate detection.)
• How can I report missing/new resources? There should be a fast and easy way to do this.
• more than one keyword should be provided seperately, i. e. not concatenated into a single string by joining the keywords with ";" or another typographic device
• keyword: fixed vocabulary might be problematic; it's more realistic to expect an open set of keywords
• A mapping across synonymous keywords in different languages would be useful.
• There is already a set of use cases (e .g. »How can I extract all NPs from a corpus?«) (note from Susanne: i. e. »Hundred questions of the F-AGs«)
• In the social sciences, it's important to ensure representative sampling from resources → complex queries for full text and metadata must be possible (also for linguistic annotations to reduce the amount of sampling errors)
• extended search would be useful