Using Prizms

Following steps install Prizms on our server.

The prereqs

Base domain: https://locv.tw.rpi.edu
Git repo: git@github.com:timrdf/locv.git
datahub.io entry: https://datahub.io/dataset/locv

Installation

Get curl:

sudo apt-get update; sudo apt-get install curl

Get Prizms:

cd; bash < <(curl -sL https://purl.org/twc/install/prizms | grep -v "^#..bin/bash$")

Install Prizms using 8 configuration parameters:

opt/prizms/bin/install.sh --me https://tw.rpi.edu/instances/TimLebo --proj-user locv --repos git@github.com:timrdf/locv.git --upstream-ckan --our-base-uri https://locv.tw.rpi.edu --our-source-id us --our-datahub-id locv

Getting started with the data: Access metadata

First, we get a copy of the repository:

git clone git@github.com:timrdf/locv.git

We can then record access metadata for the BMS-26 dataset listed at https://lmb.informatik.uni-freiburg.de/resources/datasets/moseg.en.html using the following csv2rdf4lod commands. The access metadata ends up in our GitHub repository at access.ttl and page.ttl.

bash-3.2$ cd locv/data/source
bash-3.2$ cr-dcat-retrieval-url.sh bms-26 https://lmb.informatik.uni-freiburg.de/resources/datasets/moseg_dataset.zip
Q: Make directory for source-id/dataset-id: "lmb-informatik-uni-freiburg-de/bms-26" ? [y/n] y
source/lmb-informatik-uni-freiburg-de/bms-26/access.ttl
bash-3.2$ cd lmb-informatik-uni-freiburg-de/bms-26/
bash-3.2$ cr-page.sh https://lmb.informatik.uni-freiburg.de/resources/datasets/moseg.en.html
source/lmb-informatik-uni-freiburg-de/bms-26/page.ttl
bash-3.2$ git add -f page.ttl access.ttl 
bash-3.2$ git commit -m 'bms-26 metadata'
bash-3.2$ git push

Retrieving datasets on the server

Based on the new metadata in the repository (i.e, the access.ttl file), the Prizms server will:

recognize the access metadata,
set up a new versioned dataset,
download the zip (capturing its provenance)
analyze the file formats of the zip and its contents,
extract the zip (capturing its provenance)

For example, for the dataset source/vision-mas-ecp-fr/facades-benchmark-2011, Prizms set up a new version version/2011-May-31 with a source/ directory containing the files:

cr-droid.ttl                    
ground_truth_2011.zip           monge_42.png
ground_truth_2011.zip.prov.ttl  monge_42.png.prov.ttl

source/ground_truth_2011.zip.prov.ttl contains the provenance of downloading the zip from https://vision.mas.ecp.fr:

<ground_truth_2011.zip>
   a nfo:FileDataObject, prov:Entity, pmlp:Information;
   prov:wasQuotedFrom      <https://vision.mas.ecp.fr/Personnel/teboul/files/ground_truth_2011.zip>;
   prov:qualifiedQuotation <quotation_7077d5c3-6c21-45c5-8533-d32d8678b7b3>;
.
<quotation_7077d5c3-6c21-45c5-8533-d32d8678b7b3>
   a prov:Quotation;
   prov:entity <https://vision.mas.ecp.fr/Personnel/teboul/files/ground_truth_2011.zip>;
   prov:atTime "2014-02-06T17:09:07+00:00"^^xsd:dateTime;
.

cr-droid.ttl includes file format and zip containment metadata:

@prefix dcterms: <https://purl.org/dc/terms/> .
<ground_truth_2011.zip>
   dcterms:format <https://provenanceweb.org/formats/pronom/x-fmt/263> .
   dcterms:hasPart <ground_truth_2011.zip/monge_42.png>;
.
<ground_truth_2011.zip/monge_42.png>
   dcterms:isPartOf <ground_truth_2011.zip>;
   dcterms:format   <https://provenanceweb.org/formats/pronom/fmt/11>;
.
<monge_42.png> dcterms:format <https://provenanceweb.org/formats/pronom/fmt/11> .

source/monge_42.png.prov.ttl contains the provenance of extracting monge_42.png from ground_truth_2011.zip:

<monge_42.png>
   a pmlp:Information, prov:Entity;
   prov:wasQuotedFrom <ground_truth_2011.zip>;

Publishing the dataset and its metadata, provenance

All of the provenance and metadata (in RDF) gets published as the dataset with URI https://locv.tw.rpi.edu/source/vision-mas-ecp-fr/dataset/facades-benchmark-2011/version/2011-May-31. That dataset will mention the zip file that we downloaded (e.g. ground_truth_2011.zip), and the image files that we extracted from it (e.g. monge_42.png). So, the data can be accessed by walking/querying its RDF metadata descriptions.

results of:

PREFIX foaf: <https://xmlns.com/foaf/0.1/>
PREFIX dcterms: <https://purl.org/dc/terms/>
SELECT *
WHERE {
  GRAPH <https://locv.tw.rpi.edu/source/vision-mas-ecp-fr/dataset/facades-benchmark-2011/version/2011-May-31> { 
    ?s dcterms:format ?o
  }
}

What is next?

Datasets

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Using Prizms

The prereqs

Installation

Getting started with the data: Access metadata

Retrieving datasets on the server

Publishing the dataset and its metadata, provenance

What is next?

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Clone this wiki locally