Ještě to bude chtít zobrazit počty… bodů celkem + v těch kategoriích až se bude filtrovat. Mam dvoje nová data a řekl bych že tam něco chybí ale bez tohoto to budu muset počítat jinak.

• Read UC2 in the paper
• see my latest commits to the repo
• ruian data source already added
• New analyzer: ldvm-a-ruian-obce.ttl - should be the one that extracts all "obce" from RUIAN - nested one
• New analyzer: ldvm-a-ruian-geocoder.ttl - 2 inputs, one from the previous analyzer, 2nd input will be connected to a datasource created from example-obce-ruian.ttl. Here, implementation is needed. Basically take all data from dataset input and pass it to output. Add for each entity with the "ruianlink:obec" property leading to RUIAN obec the s:geo of the "obec" from RUIAN input.
• Discovery will connect the ruian DS to the first input and then the previous analyzer after RUIAN input. The first choice will not work due to ruian size. The second one is the correct one.
• Output should be consumable by GMap visualizer and 2 points should be shown.

Such as this one: http://codemirror.net/1/sparqltest.html or another one that can be easily added.

It would help here: http://live.payola.cz:29090/components#/detail/2

Allow upload of multiple ttl (and other?) files to a single datasource.

Use case: Upload data, schema, links - as in Use Case 2.

See

http://live.payola.cz:29080/visualize/datacube#/id/32?view=chart&chartType=column&isPolar=false&p=bf042521c0e926657d5eb70aa7518b2d

Labels of dimensions are not displayed. For example, this dimension:

http://linked.opendata.cz/ontology/application/mfcrapp/rozpocty/refPeriod

has a label defined (rdfs:label), see

http://internal.opendata.cz:8890/describe/?url=http://linked.opendata.cz/ontology/application/mfcrapp/rozpocty/refPeriod

When I have a hypothetical transformer from gr:legalName to s:title, its datasample is just somethign that has s:title...I loose the expressiveness of the previous data sample. This leads me to data sample transformations - Another difference between Analyzer and Transformer - transformer does not have a output data sample, but output data sample transformation - sparql update query that transforms previous data sample. Is it a special construct? Or is it actually just execution of the transformer on the output data sample??

This is a blocker for merging the ldvm branch to master.

When I search for visualizations of dataset A and I choose to see results when combined with other datasets, show only results that actually include other datasets.

Here the datasource was 3 URLs

We should group components by their type - maybe all components with the same descriptor?

Motivation:

• Multiple implementations of SPARQL - Virtuoso, Jena, Sesame - we don't need to offer all pipelines three times, each time with a different implementation (and more times when multiple sparqls in a pipeline)
• Multiple hierarchy visualizers - SunBurst, TreeMap, ... - we need to offer visualization using one of the visualizers in a group, does not make much sense to offer a special pipeline for each of the visualizers

Possible impact:

• descriptor of a group, group membership, differentiating properties (each one in a group may have some "special" properties?)

This is something a reviewer will definitely point out - that the visualization of the pipeline is incorrect due to undirected edges and confusing because of a missing tree layout. Consider this low prio, but nice to have.

When the data cube visualizer is embedded to a space with a height smaller than the height of the side menu for setting dimension values and choosing measures, it is not visible whole and it is not possible to scroll it. Maybe adding a scrollbar to the side menu could help?

See

http://internal.opendata.cz/dataset/mfcrapp-rozpocty-a-populace-okresy-predskolni-a-zakladni-vzdelavani/resource/1c6aa5f8-99ac-49d7-8af8-2e82820c2cbe

[13/10/14 10:52:44] Jakub Klímek: ja myslel jestli to neni mezerou jako oddelovacem 2 grafu
[13/10/14 10:53:15] Jakub Klímek: skoro to ale vypada ze jo
[13/10/14 10:53:27] Jakub Klímek: protoze ted jsem udelal to samy, ale oddelil jsem to newlinem a to je ok
[13/10/14 10:53:44] Jakub Klímek: jsou to ty vizualizace FINU108 (blbne) a FINU108 (2) (jede)
[13/10/14 10:54:21] Jakub Klímek: co jsem totiz udelal je ze jsem to do novy vizualizace zkopiroval z listu tech existujicich a tam je to s tou mezerou

[13/10/14 11:26:47] Jakub Klímek: taky kouakm ze kdyz mam jednu measure a pro to to spocitam a prepnu na jinou measure tak to zobrazi data k ty stary dokud nedam refresh
[13/10/14 11:28:59] Jakub Klímek: coz je vlastne mozna konzistentni, co
[13/10/14 11:29:15] Jiří Helmich: no, on ten graf z nejakyho duvodu udela takovy to animovany prekresleni, kdyz se znovu objevi… ten refresh je nutnej, protoze kdyz mas velkou kostku, kde mas stovky hodnot na dimenzích, tak kdyby to pri kazdy prepocitalo, tak to akorat shodi virtuoso a nic
[13/10/14 11:29:21] Jiří Helmich: a tady je to jen konzistentni
[13/10/14 11:29:44] Jiří Helmich: asi zkusim nejak schovat ten graf, kdyz neodpovida vyberu…?
[13/10/14 11:29:50] Jakub Klímek: ok, chtelo by to indikovat ze data uz neodpovidaji zvolenym parametrum a ze je treba refresh

[13/10/14 11:03:01] Jakub Klímek: jeste je tam bug kdyz vic polozek dimenzi ma stejnej label - je tam pak videt jen jedna z nich
[13/10/14 11:03:08] Jakub Klímek: http://live.payola.cz:29080/visualize/datacube#/id/15?view=chart&chartType=line&isPolar=false&p=3ccdb29524a79359a6ec6fee97afdb4
[13/10/14 11:03:25] Jakub Klímek: (rozpoctovy program - jeden z nich vede na hodnotu)
[13/10/14 11:19:50] Jakub Klímek: zda se ze to dela jen kdyz je to na ose X grafu. Kdyz je to jako id řady tak je to ok

Related to #18 only for data sources and forward discovery

A user should be able to create a data source from

• Public SPARQL endpoint (URL, optionally list of graphs)
• Private SPARQL endpoint (URL, username, password, optionally list of graphs)
• File upload (optional serialization autodetection (if easy in Jena), otherwise input format, upload box)
• URL of file (optional serialization autodetection (if easy in Jena), otherwise input format, link)

Unable to move the map.

e.g. here: http://live.payola.cz:29090/components#/detail/34 it would be nice to see somewhere at the top of the page that this is actually a Data Source.

As we only have 4 types of components so far, consider using decent color coding (of the title? and later in pipeline)
e.g. Red - DS, Green - Viz, Blue - A, Yellow - T

related to #16, but now in the list.

čau, asi bug...http://live.payola.cz:29080/visualize/datacube#/id/13?view=chart&chartType=line&isPolar=false&p=8da2067c6d77405722cb5654246400a3 nejsou tam labely measures

Here it only uses RUIAN, not the uploaded file

There is a minor problem remaining. Try this:

1. http://vis.payola.cz/datacube-population-czso
2. Polar
3. switch to the second structure in the "Available data structures"

Would be great also for each dimension value, but that would be probably very slow.

Next, it would be nice if we invite some advanced sorting based on the type of data and metadata as I wrote in an email which I repeat here (in Czech):

Kdyz uz mam nejakou konkretni kostku, tak uz vim i prvky na jejich dimenzich. Delam ted zrovna treba kostku pro Ceskou spravu socialniho zabezpeceni a tam je dimenze reprezentujici kategorie duchodu (dle jejich vyse):

1-999
1000-1999
2000-2999

atd.

Tady si pripravuju generovane resources ve tvaru:

http://linked.opendata.cz/generated/resource/range/R1T999
http://linked.opendata.cz/generated/resource/range/R1000T1999

atd.

Kdokoliv muze prijit a udelat si URI libovolneho range, ktery potrebuje. Nase sluzba by mu mela vygenerovat par dat k tomu (typ, labels, atd.).

Tot ke generovani. Kdyz uz ale nejakou kostku mam, tak vim presne jake hodnoty na te dimenzi budu potrebovat, takze si vyrabim jejich TTL reprezentaci a tu pripojim k datum kostky. A neni problem, abych si je rovnou i usporadal. A to usporadani bych chtel reflektovat ve vizualizaci.

Pokud to usporadani dodano nebude, jak pise Jirka, tak pak musime mit nejake vestavene usporadani.

Jeste me napada, ze bychom mohli k definici dimenze pridat propertu, ktera by rikala, jak se ma usporadavat (alfanumericky, numericky, ...).

Posilam priklad, at je to jasnejsi:

SITUACE: CSSZ ma nekolik dimenzi, kde pouziva celociselne rozsahy typu 2000-2999. Ma nekolik dimenzi, kazde pouzivaji trochu jine hodnoty. Udelal jsem si SKOS concept scheme, ktery obsahuje vsechny mozne rozsahy, ktere CSSZ pouziva:

http://opendata.cz/martin/data/czlod/concept-scheme-ranges.ttl

Kazda konkretni dimenze, ktera se muze objevit v kostkach CSSZ pouziva podmnozinu techto rozsahu.

Premyslim, jak pro danou konkretni dimenzi rici, jak se ma usporadavat. Napada me:

1. Pomoci skos:OrderedCollection - pod tim odkazem priklad pro jednu dimenzi (resource ran:ordering1). Jedna se o standardni zpusob usporadavani resources. Jsem schopen ho udelat, protoze to delam pro konkretni dimenzi, jejiz (konecnou) mnozinu hodnot dopredu znam. Ten ordering bych jeste nejak potreboval navazat na definici te dane dimenze nebo primo na ran-onto:RangeScheme, coz nevim, jeslti existuje neco standardniho. Property, pomoci ktery bychom navazali dimenzi/conceptscheme na ordering bychom asi museli zavest, nebo nejaka existuje?

2. Stejne jako v 1) udelat property, ktera specifikuje ordering pro dimenzi nebo rovnou conceptscheme. Misto usporadaneho vyctu hodnot jako v 1) bychom ale dali URI zpusobu usporadani.

As discussed on Skype

1. Schedule discovery starting from all data sources (for starters, once every 2 hours)
2. Flag the result with a special flag
3. Add links from visualizers "What can I see with this?" (list and detail) leading to a list of pipelines ending by this visualizer in the latest flagged discovery

For example, here

http://live.payola.cz:29080/visualize/datacube#/id/32?view=chart&chartType=column&isPolar=false&p=bf042521c0e926657d5eb70aa7518b2d

it is hard to understand what is displayed. The problem is that the chosen measure and dimensions fixed to a single values are not shown somehow in the visualization.

I propose to show (e.g. under the visualized graph) the measures and dimensions which are not visible in the following form:

• show the title of the data cube (rdfs:label, dcterms:title) as the main title (currently there is fixed "DataCube"
• show the names of chosen measures under the main title as secondary titles
• show pairs [Dimension name] : [Chosen value] at the bottom of the graph for each dimension which is not visualized in the graph

See example:

Moving a dimension 1 position down or up does not move the dimension below the following or above the previous but only does +1/-1 in the ordering. As the result, when I start with

Dim X, position 1
Dim Y, position 2

And want to move X one position below, I can see

Dim Y, position 2
Dim X, position 2

I.e. both are on the same position.

Reported by @martinnec

TreeMap or other hierarchy visualizer? Bug in data?

Add zoom buttons.

Here, it has live.payola.cz label

http://bl.ocks.org/mbostock/5944371

Discovery detail leads to e.g. http://ldvm.opendata.cz:29090/pipelines#/list/?discoveryId=3 but that still puts me on the pipeline list. I think I should see the discovery process log.

Try to add GA events for various actions so that we can monitor user behavior

Link to pipelines that end with the visualizer

Via a URL, it should be possible to specify, which values on which dimensions are active.