Anthony's Blog

Webfinancial site (Vershon's Blog) Hey Tony, can you let me know what the new address for the interface is so I can put it in to the entry page.

Vershon, use the following address for the interface;

http://people.emich.edu/ataylo15/webfinancial.html

It currently doesn't exist but will shortly.  Since the table visibility wasn't created by this past Monday; I created a polymorphic function that dynamically creates a table and displays information based on what arguments are passed to it.  It's pretty sweet!

I don't know what's up with the menu bar; if something isn't posted or emailed by tonight I'll do that too.

Davidia, I've provided a couple links that graphically depict what you describe in your blog.

XBRL (Davida's Blog) XBRL consists of an instance document, containing primarily the business facts being reported

XBRL (Davida's Blog) Taxonomies are a collection of XML schema documents and XML documents called linkbases by virtue of their use of XLink.

It's kind of interesting after having worked with instance documents for a little while.  I think that the way that XBRL is formatted is kind of silly in some regards.  Referring to different parts of a single document is messy.  Loading into and processing in a web interface is difficult because of this.   An attribute for each tag that describes the context is equally beneficial-- AND you wouldn't have to refer to different sections of one document at the same time.

I haven't come across any documentation that explains why context ids and references are set up like they are.  Most likely, it's beyond the scope of this project.

Context parsing (Ryan's Blog) Ok I've tried to implement how you had the sample xml into the engine. Hopefully this will make things a little easier for parsing.

Ryan, that's terrific!  I was wondering if you could do one more thing.  Will it be possible to open the flood gates and let ALL of the data flow through; in this format?  I have all of the required tags coded in JavaScript so I can still compute the different ratios.

I guess it depends on how your pulling the tags.  If your pulling the data based on the tag names, we might have a problem letting everything flow through.  If you're pulling based on context ids and references, it should be okay, right?

I really want to get all of the tags for additional information display.

What you have done is absolutely terrific!  I really appreciate everything that you have done!

Re: Ah yes; now I see what you mean (Ryan's Blog) Also I don't know if I want to make large changes to the xml structure this late because then we also have to make changes to the javascript to load it. Maybe if you could write some xml in the format you want and I could understand it better and see what you mean.

I agree: large changes to the xml structure this late is not a good idea.  I'm also providing a sample based on 3M's XBRL document so you can see what I meant.

I don't know how much effort it would take to accomplish this.  If it's in your best judgment to not change the structure; I trust you.

The most important thing is to group together the data by date.  The start date and end date nodes separate needed elements.  In this example, all of the child nodes under date 09/30/2006 and date 07/01/2006 in fact belong to the same period, 07/01/2006 - 09/30/2006.  There isn't a reference provided so I am unable to display these elements to the interface.

I could try using JavaScript to relate the two.  Afterall, the server script does return the url for each company's instance document...

Also, please don't rename any of the tags...  The different (yet similar) tags that are identified are coded into JavaScript.

Thank you for considering.

Re: Webfinancial data (Ryan's Blog) I believe what is happening is that I used the start date for the filing date which is the one in July there. If we need that to change to the end date, or forward both dates through or something that can be done.

I see what you mean...  For that particular company and particular form; the examples provided had a context reference of something like "ThreeMonthsEnded_30Sep2006".  The server script separates the data into two nodes, the start node and the end node.

Is there a way to put these two 'filing' nodes together?

And if I could give it one more shot... : )  If there was some way to output the results by context-- we would have every tagged item for every company available to work with.  This would be awesome because (of both of our headaches) some companies use different tags for the same thing.

i.e. cashcashequivalents vs. cashcashequivalentsshortterminvestments.

Maybe different (technically) but the same when computing cash ratio (unless they use the tag marketablesecurities).

The different tag names makes it time consuming to search for every possible variation of tags through the monstrous documentation for the nearly 2 dozen namespaces.  Not to mention, XBRL (a.k.a. XML) is extendable and some companies create THEIR OWN namespaces.  Headaches turning into migraines quickly.

Is there anyway to make it easier???

I'm providing an updated link to show what I've done with the javascript code for the project.  I made sure to provide lots of comments so you can see what's going on.

Update (Ryan's Blog) As always, let me know if you need anything else forwarded through.

Well, actually I was wondering if it would be possible to change the way that we grab data from the instance documents.  I was reconciling our interface output, the php output, and xbrl instance doc and found some discrepancies between them.  In these examples, the php output doesn't provide net income for Sept 30, 2006 even though it's provided in the instance doc.  Instead it places the net income (894000000) under filing date 07/01/2006.

Would it be possible to grab the data based on context?

Examples
Bristol Myers Squibb context id and context reference
Altria Group context id and context reference

In both examples, the respective company uses their own context id but always references the context with the data.  If it were possible to output (using php) by context, providing the start date, end date, and all tags referencing the context id; it would be a lot easier to do what we are trying to do.

You have done a lot with the server script and I would understand if it was too much work given the date of iteration 3.  But I feel that it would probably be easier to solve our data issue this way.

PHP & JSON (Ryan's Blog) If anyone wanted to look into some server magic with JSON instead of XML.

Using XML is fine; I wrote a constructor function that defines what company data is passed back from your php script.  I then load the 'company' objects in an array.  I'm going to use the array to compare company names and populate the drop-down lists (unique company name, formtype, and filing dates).  It works fine in both FF and IE.

I'm going to clean up the JavaScript code and post my progress later.

Thanks for updating the server script, looks great!!!

http://people.emich.edu/ataylo15/is449/project/powerpoint-11.27.06.ppt

 

eventmaps groupproject prototype2 review (Kokila's Blog) Great to hear that calendar control works in firefox 2.0...

ScreenShot

I wonder why it doesn't work in v1.5???