TWiki> Astrogrid Web>AgScience>AGInterfaceMeetingMay2007 (2007-06-04, NicholasWalton)EditAttach

AstroGrid Interface Meeting: Friday 25th May 2007 at UCL

Notes from meeting

These notes (written by Andy just after the meeting) summarise the key ideas that came out of the interface meeting. By the end of the day these ideas were a firm consensus amongst the people present but they still need wider discussion. They will be taken forward for discussion at the planning meetings and then the consortium meeting before firm decisions are taken.

  • The top priority is, within a year, finishing stuff that works, runs quickly, looks professional, and is clearly something you can't do more easily simply by going to a data centre web page. This may mean cutting back on aspirations to be sure of delivery.

  • There are three main entry points into the VO world - through application specific tools, through data centre web pages, and through "VO portals". A large part of the work of AstroGrid is building the infrastructure for tools writers and data centres to use. We shouldn't try to duplicate what they do - they do it better and we don't have the resources. However we don't just leave them to it - we need to work with them directly

  • We must also offer some kind of "AstroGrid portal". This entails at least some simple tools, to reach the point of actually delivering data into science applications. We should limit these to a core set that we can confidently deliver over the coming year. The "AstroGrid portal" should be just ambitious enough but not too ambitious, and should stop short of what data centres and tools writers do better.

  • It was uniformly agreed that VO explorer is a radical improvement, and a kind of paradigm shift in approaching the VO. It was felt that VO Explorer should replace Workbench as the "AstroGrid portal" as soon as it is ready and stable.

  • Users will still want sometimes to launch one-off components, such as browsing MySpace. It was agreed these should look like any other application on your PC - for example on Windows, appearing on the "Start" menu under a VO Program Group.

  • We did not agree with the idea of a "naked AR" launch; if VO Explorer is the portal, we should continue to carry AR inside this.

  • VO Explorer is already good but still has some way to go in useability terms. There was a general feeling that as much complexity as possible should be hidden from the user (at least at first). This was not so much to do with the basic layout and logic, but more to do with the bewildering metadata choices and what they meant. (See Registry bit later)

  • People liked the basic logic of (a) find resources (b) take action to find data (c) hand over data to third party apps. Within (b) there are the integrated tools. It was agreed that these should be astroscope, task launcher, and query builder, with some additional interest in a crossmatch tool (see below).

  • It was agreed that having the "generate script" type actions within the integrated tools is going a step too far, at least in the basic version. Possibly an "advanced version" would reveal these later.

  • We looked at Taverna, and discussed the whole idea of workflow. Taverna still looked quite intimidating, and it wasn't clear what it did that scripting couldn't, with the possible exception of making parallelism easier. There was much discussion of whether astronomers in future would more easily take to a graphical interface or a scripting interface. But for now the feeling was to let the Taverna story develop slowly, and to put our effort into providing scripting. It was thought more important to extend scripting to IDL (and perhaps Perl) than to work on graphical workflow.

  • For scripting, we did not see having any kind of GUI interface as a priority for now. Rather, we should continue the current work on documenting the methods and giving examples.

  • Astroscope. There is an existing list of fixes and improvements that should be implemented - but not of course the things that are now provided by the VO Explorer front end, such as restricting wavelength etc. A very desirable addition is the ability to run queries on an uploaded list of positions. However, this may be better as part of a crossmatch tool (see below).

  • Task Launcher. A template that is customisable by the applications provider is probably the way of the future, but in the short term there is a need for a generic interface to parameter driven applications. However this is currently very user unfriendly. A concentrated design mini-project with a scientist and an engineer should fix this.

  • Query Builder. The subject of much discussion. Current query builder much improved but somehow still not quite right. The web-page wizards used by SDSS and WSA etc are much more intuitive. It was felt that addressing single-table catalogues such as those in Vizier should be a much easier problem to solve than building a generic query builder for multi-table relational databases. We did not come to firm conclusions in this area, but resolved to urgently investigate the problem further.
    • [naw: it is not at all clear that the web based wizards are useful. The vizier query page might be a better example. Also remember that there is a new query builder about to be deployed which is much better than the xcurrent one, thus this should be tested first before we move further in this area. This awaits voexplorer deployment thoughj]

It was noted that some resources offer both an ADQL interface and a cone-search interface. Possibly they should appear as a single entry with two actions offered - "simple query" and "full query".

Providing a simple interface to all those Vizier tables would be very good.

  • Crossmatch. If you ask a random astronomer "what is the VO going to do ?" she would answer (a) make it easier to find stuff, and (b) make it easier to cross match any two catalogues. So VO Explorer finally starts to do (a) properly, and its time we cracked (b). Answer (i) is get the two tables you need into Topcat, and use its crossmatch tool. Answer (ii) is write a script using the STILTS library. However, for simple xmatches, we really want (iii) simple GUI specifying resource A, resource B, xmatch radius etc, with returned VOTable. This is a better place to do the "upload list" thing than in Astroscope - its really "crossmatch my list with resources A and B". This is an additional piece of work, but a fairly high priority.
    • AMSR 20070529 Much of STILTS cross-matching is already available in a dialogue box via task launcher, both for time-domain work (search for Cross and Match) and as a wrapped part of STILTS (search for STILTS) but the interface needs translating into astronomer-speak (and possibly the way that the functions are presented needs tweaking). Maybe Silvia and me (or some astro) should look at this with Mark T.
    • NAW one option we discussed was a crossmatch script - or even something in taverna.

  • It becomes clearer that there are a variety of problems with the registry entries themselves - titles can be unhelpful, metadata is missing, multiple entries appear under the same "Title", and conversely, the same "publisher" uses several different names - eg WFAU Edinburgh, Wide Field Astronomy Unit Edinburgh, WFAU at ROE, etc.

  • For registry entries in the UK, we can and should go on a campaign of fixing these issues ourselves very soon. For the rest of the world, this will take much longer.

  • Many of the problems can be fixed by a suitable "translation layer" in VO Explorer, and this must be a key task in VO Explorer improvement. For example, a search on "IPHAS" produces four resources all called "IPHAS catalogue". One of these is a cone search interface, one is an ADQL interface run through DSA, one is a background CEA service making the latter possible, and one is just a link to the catalogs. These should really appear as one item, and/or carefully distinguished in nature. Likewise there are two items called "IPHAS images", one of which is a straight SIAP interface, and one of which is a CEA application offering the same thing.

  • The order of priority for these Registry problems is (i) do the translation layer (ii) hand fix UK registry entries (iii) worry about the rest of the world later.
    • [rgm; e.g. in the short term we may have to 'filter' some of the 'important' world resources in order to ease of use, e.g. a search for 2MASS or SDSS should return easily the VO ready entry points to these datasets; this will hopefully be a short term fix; e.g. the ADQL interface to 2MASS is via the DSA; maybe there are others too.]

  • So the key jobs we saw coming up are (1) finish VO Explorer, including "translation layer" (2) fix up Registry entries (3) complete scripting documentation (4) implement astroscope improvements (5) re-design task launcher (6) scoping study for query builder (7) build crossmatch GUI and (8) continue the taverna investigations at the current level.

  • the aim is to have a demonstrateable version for AGOC on June 20th, and a release version around October.

Meeting Agenda

  • 1400-1600 : future

  • Overview of technical components and possibilities (KN)
  • Options for next cycle (All, chaired by AL)
  • Long term nature of UI (All, chaired by AL)

Discussion Topics

People

Apologies

Location Details

  • University College London
  • Malet Place Engineering Room 1.02
    • This is an 80 seat lecture theatre (best I could get with wireless I'm afraid)
    • QuicktimeVR of theatre
    • Getting to the room
      • Malet Place Engineering is in grid reference Q13 of the following map - click here
      • If you are getting the tube, I recommend going to Goodge Street (on the Charing Cross branch of the Northern Line) and walking down Torrington Place (off Tottenham Court road)
      • The building is glass fronted with sliding doors.
      • On entering the building, take the stairs (to the right) to the first floor (the lifts apparently don't stop there). To the right of the lift doors on the first floor is a single dark grey door. Go through it and go down the small set of stairs to your right. 1.02 is the the first door on your left.
      • I'll try and remember to put some strategically placed signs up on the first floor.
  • Tea and Coffee
    • Unfortunately due to the nature of the room I have been unable to book catering for the meeting. I therefore suggest you buy your drinks before coming to the room. Below I have made a few suggestions of where to buy a tea/coffee.
      • Housman room - This may be your easiest bet if you are familiar with it. It has a vending machine which runs all day - 50p per drink. You'll find it in grid reference F2.
      • UCL Shop - located at the South Junction (grid reference F4, you can't miss it), doesn't have tea or coffee but it does have a variety of snacks and sandwiches.
      • Waterstones Gower Street - Waterstones is very close to Malet Engineering and I believe there is a Costa Coffee in the basement.
      • LunchBox - This coffee shop is on the ground floor of the University of London Union which is itself, across the road from Waterstones.
      • Lower Refectory - This is on the UCL campus. You will find it in grid reference G3.
      • Tottenham Court Road - There are a large number of coffee shops along Tottenham Court Road representing all chains possible.
  • Lunch
    • As above, I haven't been able to arrange for lunch for the room. All of the options above provide lunch in different ways. The Housman room serves lunch from 12-2pm and has hot food and sandwiches. The Lower Refectory also has a larger range of hot foods and sandwiches but is not exclusive to staff. There are several pubs nearby that serve hot food including the Marlborough Arms and the Bar at TCR although I would expect them to be very busy.
  • Wireless
    • I have received the information for the wireless accounts, this will be distributed on the day.
  • If you have any problems finding it on the day then call me (Dugan) on my mobile (07840-0750896)

-- DuganWitherick - 11 May 2007

Edit | Attach | Print version | History: r18 < r17 < r16 < r15 < r14 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r18 - 2007-06-04 - 14:08:55 - NicholasWalton
 
AstroGrid Service Click here for the
AstroGrid Service Web		 This is the AstroGrid
Development Wiki

This site is powered by the TWiki collaboration platformCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback