Tailoring Information Presentation with Words and Pictures at the COSSC

James M. Jordan, Senior Systems Engineer
Mark Cresitello-Dittmar, Programmer/Analyst
Jesse Allen, Programmer/Analyst
Hughes/STX

Abstract

Hypertext's advent on the internet has allowed the Compton Observatory Science Support Center (COSSC) to explore new ways of presenting its archived data to the scientific community. Until recently, Compton Observatory data could only be accessed using anonymous ftp and a vt100 interface with a special account. Now, via the web, thumbnails of sky images are linked directly to the data, while forms and imagemaps allow data to be selected based on a wide range of characteristics.


  1. Introduction
  2. Where We Were
  3. First Attempts at Data Presentation
  4. Breaking into Forms and Imagemaps
  5. Conclusions
  6. Authors

1. Introduction

Making the public gamma ray data from the Compton Observatory satellite available to the astronomical community is the primary mission of the Compton Observatory Science Support Center (COSSC). With the advent of hypertext, we have been given the opportunity to explore various ways of presenting and interlinking the available data in order to make it more easily accessible than the dry, and uninformative, listing of the data files obtained using anonymous ftp.

Our first attempts to use hypertext to go beyond anonymous ftp include the use of a table with the table entries linked to their corresponding data files. We also used pages with thumbnails of sky images which link the user to the data which the images map. Each of these methods of presentation has the benefit of providing some additional information beyond the name of the files which are available in the archive. The additional information assists in finding useful data. But in each case, the user is still presented with a large amount of information to process in making a selection.

Our current goal in data presentation is to develop methods of allowing the user to guide filtering mechanisms for selecting data. Two methods are being worked on. The first allows the selection of gamma ray bursts based on user selectable parameters in an html form. The second combines a form and an imagemap to allow the user to select objects from a map of the high energy gamma ray sky and specify which types of objects are of interest and the resolution of the selection on the sky map.

2. Where We Were

Data at the COSSC is stored in a set of directories accessable via anonymous ftp. Four instruments, each producing a variety of different data types, are on board the Compton Observatory satellite. Many of these data types currently reside in the archive, and more will in the future. Before the availability of the web , our facilities to access the data were limited to anonymous ftp and an ASCII database forms interface.

To ease anonymous ftp access, the data directories are divided up by instrument and then by data type and/or viewing period. So, for instance, if a guest investigator (GI) wanted to look at burst number 110 detected by the BATSE instrument, simply following the directory tree is all that is required to get to the data. First, "cd" to the BATSE directory, then to the bursts directory, and then to the BURST_00110 directory. Simple, obvious, and intuitive! But, only if you know exactly what you're looking for. Generally, this is not a good premise for we, the archive maintainers, to take.

Going a step further then, as the archive is cataloged in an Ingres database, we built a small vt100 Ingres forms interface to assist the GI in selecting information in the COSSC archive. Through the interface, a GI is able to specify a time period, an object, and even an instrument of interest. The query is then started and, soon, a list of file sets is returned. Although the file set names are somewhat informative, they're still rather reminiscent of the listing of filenames that is obtained from anonymous ftp. The GI is able to winnow the file list to some extent, but it is still, generally, very long and gives no indication of the parameters of the data stored in the files.

So, here again, the added value in data selection is rather limited. The limitations stem primarily from the lack of an interface powerful enough to display the parameters of the data efficiently and clearly for a GI to use in data selection.

3. First Attempts at Data Presentation

Then came the web and Mosaic. The combination of the two seemed to provide a way of displaying a lot of data visually in such a way as to give a GI some idea of the parameters of the data available for selection. Initially, our thoughts again turned to displaying all the data for perusal by a GI and then linking the display back to the data itself.

Displaying thumbnails of sky images from the EGRET instrument was the first foray into data presentation on the web(EGRET Page). A GI is first shown a dynamically created list of viewing period ranges. Below is an example viewing period list:


EGRET Pointings

Clicking on any of the selections below will present a set of six glimpses of EGRET data. Each glimpse of data can then be clicked on for more detailed information on the data, to access a gif of the data, and to get the data itself.

EGRET Pointings 0.2 to 0.7

EGRET Pointings 1.0 to 5.0

EGRET Pointings 6.0 to 9.5

EGRET Pointings 10.0 to 14.0

EGRET Pointings 15.0 to 20.0

EGRET Pointings 21.0 to 25.0

EGRET Pointings 26.0 to 31.0

EGRET Pointings 32.0 to 36.5

EGRET Pointings 37.0 to 42.0

EGRET Pointings 43.0 to 43.0


Each range list is an anchor which creates a page of six EGRET sky images. The images, combined with the location information printed with the image, provide the user with some idea of the nature of the data obtained during each viewing period of the EGRET instrument. Each image is also an anchor which can be selected to access data from the EGRET instrument. A page of thumbnails looks like this:

EGRET Pointings

Go to previous image set

Pointing 1.0        Pointing 2.0        Pointing 2.5       
RA    5h52.28       RA    20h5.56       RA    5h51.32
Dec     17.14       Dec     36.58       Dec     12.47
L      190.92       L       73.28       L      194.86
B       -4.74       B        2.56       B       -7.29

Pointing 3.0        Pointing 4.0        Pointing 5.0       
RA   12h46.16       RA   11h59.36       RA    18h1.56
Dec      2.62       Dec     41.52       Dec    -30.96
L      299.76       L      156.18       L        0.00
B       65.46       B       72.08       B       -4.00

Go to next image set


Certainly, the images are at least an aesthetic improvement on the file list, but the GI is still being presented with a large list of data with little to distinguish it. Although the positional information is shown with the image, certainly, at least having some further numerical data accompanying the pictures and describing the image might be useful. And even then, the user would still have to browse the entire selection of data to ensure finding all of the data of interest.

Another area where we present GIs with all available data and tried to use the web to make some sense of that data is with the BATSE instrument's burst detections(BATSE Bursts Page). The BATSE bursts are listed in a table with position and date information. Here's a small portion of the table:


BATSE Burst Table

BASIC TABLE

There are twelve columns in this file:
  1. The BATSE trigger number.
  2. The BATSE Catalog burst name.
  3. The truncated Julian Date (TJD) of the trigger TJD = JD - 2440000.5
  4. The time in decimal seconds of day (UT) of the trigger.
  5. right ascension (J2000) in decimal degrees.
  6. declination (J2000) in decimal degrees.
  7. Galactic longitude in decimal degrees.
  8. Galactic latitude in decimal degrees.
  9. radius in decimal degrees of positional error box.
  10. angle in decimal degrees of geocenter (the angle between the burst and the nadir, as measured from the satellite).
  11. overwrite flag: Y(true) if this burst overwrote an earlier, weaker trigger. N(false) otherwise.
  12. overwritten flag: Y(true) if this burst was overwritten by a later, more intense trigger. N(false) otherwise.

Burst Table


   105 2B 910421    8367   33243.789     269.300      26.500      52.040      22.970       0.560     121.900  N  N
   107 2B 910423    8369   71684.750     196.700      -5.200     310.060      57.460      13.880     146.600  N  N
   108 2B 910424    8370   71006.602     171.900     -24.900     279.840      34.180      12.140      62.700  N  N
   109 2B 910425    8371    2265.740      92.400     -18.600     225.360     -17.370       0.960      84.300  N  N
   110 2B 910425B   8371   20253.320     346.900      22.600      93.640     -34.300       6.360      94.000  N  N


Again, as with the display of the EGRET thumbnail images, there is some added value to the approach of displaying the table of all the BATSE bursts and having links to further information and to the data itself. Certainly, the presentation of the date information allows a GI to select only those bursts in a range of interest, but the vt100 Ingres forms interface did that automatically. The position information may help another GI who is interested in data from a particular location of the sky, but there's no facility for automatically selecting on the position numbers, nor on the numbers and information presented in the pages of further information to which the burst table is linked.

For example, if a GI instead wanted all of the bursts detected by BATSE with a duration greater 50 seconds, then getting that information from the listing of burst numbers is quite possible. Each burst can be selected individually, then the for each burst the full report must be selected, and then the information must be extracted from each full report. Added value in that the information can be had online, but through a tedious and laborious process of information dredging.

Both of these cases, the EGRET sky thumbnails and the BATSE burst table, add some value to the data scanning process done by GIs. But, the GI still has to scan all of the data in order to ensure that all appropriate data has been found and to get some feel for the overall characteristics of the data. Additionally, the data which is available online is either limited, in the EGRET case, or presented in a format which can take some time to digest, in the BATSE full data report.

4. Breaking into Forms and Imagemaps

To further tailor the presentation of the data archived at the COSSC to a GIs need, we started working with HTML forms and Imagemaps. Ease of data selection and the palatable presentation of a significant amount of data are the two areas which seemed to need the most immediate addressing. Enhancing each of these areas also was a good way to add value to the selection of the EGRET and BATSE data types buy cutting down the amount of time spent by the GI selecting data.

Getting the server computer to do the searching through the BATSE data is an obvious first step in adding value to the data presentations at the COSSC. Even if the server were just to search through the data in the basic table, as shown in the example, at least the GI would have fewer bursts to look through if some paring down of the data could be done based on date and burst position. But there is much more interesting data available describing bursts. Beyond the basic table from the example, there are several other tables which present various burst parameters. Certainly, each of these tables could be put online in the same fashion as the basic table to allow the GI scan the information and then select bursts for a review of the further hyperlinked information. But the GI is still scanning an enormous amount of data, and this time, the data is spread over several tables.

So we created an interface which has the user select parameters of interest, give bounds to those parameters, and then let the computer do the drudge work of scanning all the tables and presenting a combined table of the selected information(BATSE Burst Selector). The interface, seen below, presents the GI with a number of parameters to select as being active for a data search or not, and then allows the GI to select what sort of range will be imposed on those parameters.


Select Batse Bursts

Plot all of the bursts in the catalog
Select the values of the following parameters and then submit the request with the Execute button, an Aitoff projection of the selected bursts and links to the burst data will be returned (Form Help):

Duration (seconds -- typical value = 15)
    T50
    T90

Cmaxmin (Cmax/Cmin -- typical value = 3)
    64 ms
    256 ms
    1024 ms

Peak Flux (photons/cm^2/sec -- typical value = 2)
    64 ms
    256 ms
    1024 ms

Fluence (ergs/cm^2 -- typical value = 6.0e-07)
    Channel 1
    Channel 2
    Channel 3
    Channel 4

After a selection of parameters is made and values are given to those parameters, then a search is made. The server scans through the tables appropriate to the parameters selected by the user, tests each entry for being within the specified user specified range, and then returns a table of the values of the parameters selected for each burst which matched all the GI's specifications. The search script also makes a nice plot of the bursts selected. For a search where the specified parameters are T50 > 50 seconds and Cmaxmin > 4, the results look like this:

5 BATSE Bursts Selected

Burst T50     Cmaxmin 64 
222   57.024  5.523      
1288  100.224 6.350      
1533  85.376  5.663      
1606  53.376  9.997      
2156  92.032  17.156     
Return to Selection Page
So, a GI can select on wide variety of parameters and then scan a much more limited selection for what is really desired. Additionally, only the data specified by the user as interesting is presented, so that there is less wading through information to just grab the desired pieces. Last, there is some added value in the ability to get information about the selectable parameters with hot linked explanations. We find significant added value in this approach to searching the BATSE data.

For looking at data from the EGRET instrument, although a similar approach to that used for the BATSE instrument was possible given the multiple tables of EGRET data, because of the characteristics of the data and of the instrument we chose to use an imagemap built into an HTML form. The BATSE instrument on the Compton Observatory looks in all directions at once. On the other hand, the EGRET instrument is pointed at and only looks at a particular region of the sky at any one time. Additionally, the EGRET data tables, unlike the BATSE data tables which are separated by data parameter, are separated by object type. That is, quasars are in one table while pulsars are in another.

Our approach for EGRET is to present the GI with a map of the sky as seen by the EGRET instrument, showing the various objects that have been seen(Egret Skymap Selector). With the form beneath the graphic, a GI can specify what types of objects are of interest and what range of the sky to scan. The GI then clicks on the map at the center of the area of interest and a new page is returned with all of the data in that region. If a GI has specified no objects, then the nearest object is selected. Here's the initial page:


Select Egret Data

Select the objects of interest with the form below the map, then select a location on the map, or, just select a source on the map:

Detected Pulsars
Strongly Detected Unidentified Sources within 10 degrees of the galactic plane
Marginally Detected Unidentified Sources within 10 degrees of the galactic plane
Positive Detections of Radio Loud Quasars and BL Lac Objects
Marginal Detections of Radio Loud Quasars and BL Lac Objects
Strongly Detected Unidentified Sources more than 10 degrees from the galactic plane
Marginally Detected Unidentified Sources more than 10 degrees from the galactic plane
Large Magellanic Cloud


The output from a search on the EGRET data, in addition to plotting the objects selected, takes full advantage of hypertext in the table of data returned. Column headers are hyperlinked to explanations when appropriate. Notes specified in the table are also linked to, and each line in the table is linked to a page with the EGRET skymap images from our original work on web access to EGRET data. The tables returned to the GI by the search of EGRET data look like this:

Selected EGRET Data


Selected galactic location:
l = 273.541504
b = -2.008971
Selected sensitivity angle:
radius = 30 (Default)

Table 2: Detected Pulsars

Source ID and Characteristics Pulsed Flux EGRET position ID Freq 1st Deriv 2nd Deriv 10^-6 cm^-2s^-1 l b error Ref Other Name B0833-45 11.20 -1.56E-11 5.65E-24 7.8+/-1.0 263.52 -2.78 0.04 degree 5,2 Vela
VP = 0.7 Offset = 4.495249 degrees from observation center VP = 6.0 Offset = 29.890953 degrees from observation center VP = 8.0 Offset = 2.947890 degrees from observation center VP = 14.0 Offset = 21.602436 degrees from observation center B1055-52 5.07 -1.5E-13 - 0.24+/-0.04 286.1 6.4 0.30 degree 7 - VP = 0.7 Offset = 20.529377 degrees from observation center VP = 8.0 Offset = 26.078213 degrees from observation center VP = 12.0 Offset = 28.529850 degrees from observation center VP = 14.0 Offset = 7.215940 degrees from observation center VP = 32.0 Offset = 16.587111 degrees from observation center

Table 4a: Strongly Detected Unidentified Sources within 10 degrees of the galactic plane

Position Flux Source ID l b 10^-7 cm^-2s^-1 VP Notes Ref 3 - 14.4+/-2.4 6.0 10.4+/-1.9 8.0 7.1+/-1.1 14.0 <22.1 32.0 9.0+/-0.8 Allsky 4 - 8.8+/-1.9 8.0 5.5+/-1.0 14.0 <11.1 32.0 5.8+/-0.7 Allsky GRO J1110-60 290.99 -0.23 51 <14.9 0.7 - - <7.1 8.0 5.4+/-1.1 14.0 <12.0 32.0 4.8+/-0.7 Allsky

Again, we find significant added value in this new presentation of the EGRET data. A GI can still access the EGRET thumbnail sky images and the data as in the previous web work on EGRET. Now though, the GI is getting the data after having seen some of the parameters which describe the data. Hopefully, sufficient parameters of the data are presented to allow a GI to easily find just the objects and data which are of interest.

5. Conclusions

We at the COSSC have found that the advent of the capabilities of the world wide web have given us tools which greatly further our mission of making data from the Compton Observatory readily available to the astronomical community. Although we were able to make the data available with anonymous ftp and a vt100 interface, the interfaces are not well tailored to giving a user a picture of the data in order to allow the user to fine tune their data selection.

Using the web initially, we were able to present the data in the COSSC archive with some information which provided a little bit of guidance for making a selection from the data. Still, the presentation was tailored to present all, or at least a large chunk of the data to the a GI. This approach still presented a GI with a barrage of data to wade through to discover appropriate data.

Finally, by more fully utilizing the capabilities of HTML and the web, we were able to tailor our presentation of the data to the data. Thus, with data from the BATSE instrument, we are able to shape the data selection mechanism around the fact that BATSE data is primarily interesting for the characteristics of bursts rather than the pointing position or the of the instrument. EGRET data also benefits from the HTML capabilities with the presentation of an all sky map which lets a GI know which objects EGRET has seen and easily get to the data about that object. In each case, using features of the web and HTML, we were able to better tailor our information presentation to the needs of our user community.


Authors

James Jordan, Mark Cresitello-Dittmar, and Jesse Allen are employees of HSTX working at NASA's Goddard Space Flight Center. They are software engineers working at the Compton Observatory Science Support Center which archives and provides access to gamma ray data from the Compton Observatory satellite, and at the HEASARC, the High Energy Astrophysics Science Archive Research Center.


Contact Author Email Address: jmj@enemy.gsfc.nasa.gov