Widgets - LIVE!

Our widgets are up on our partner's site: OptionsZone.com!!!

The first one (Question of The Day) pulls questions their experts answered.
The second one (Knowledge Share) displays a question they ask and collects answers from users.

TickerHound Update

We released a bunch of features that we've been working on for the past few weeks early this morning - after a late night coding and bug hunting fixing session.

These features are under the Tools umbrella:
Dictionary - look up 6000+ stock related terms (from investorwords.com)
Stocks - look up stock quotes, add stocks to questions (from quotemedia.com), done with Eric and Burak's help
People Search - look up users
Articles - this isn't new but we re-did some of the backend
New Search - search everything! with AJAX autocomplete!

Also, with the Monique's help (she knows how to make things pretty), we shift a couple of things around to make the site much easier to use and aesthetically pleasing. The process was brutal - Monique and Wayne made me shift pixels for hours! PIXELS!! HOURS!!

It feels like the battle is won but the war is not over... damn you IE6! I should be doing victory laps right now, not worrying about how fixes and hacks. Where do we send Microsoft the invoice for the hours we spent fixing their crappy browser?

All in all, I'm pretty proud of our accomplishments. I hate it and love it every time we release features.

Google Custom Search Engine - Part 2: What's What is

Part 2: What's What is

Back in part 1, I ended with Wayne mentioning "mad bugs". Since it was only a text message and not a phone call, I still had time to jump on the bus and head down to the office.
I need to get into the nature of our site and our search strategy before I can get into the bugs. You'll understand why after the explanation.
Tickerhound.com is a Question and Answer site based on finance questions. We launched with a couple of thousands of questions AND answers from one of our partners, Investorwords.com (they're the questions asked and answered by StaffQuestion and StaffAnswer. The Answers also give credit to Investorwords.com ).
This way, the site will have a wealth of information and be more useful for the first wave of users at launch.

The bug: when users type in "what's" (apostrophe s) as part of the question, questions that start with "what is" doesn't show up in the results. As you can imagine, a lot of the questions start with "what is" - no good.

Getting there: After more testing and probing, Wayne the MasterBeta.... tester, found than using "what is" in the query will pick up questions with "what's".
We did more testing with "what is" as the baseline for good results.
I tried doing "what's OR what is" then the question as a query but got bad results.
The next one I tried was both the "what's" and "what is" version of the question, like "what's blah blah OR what is blah blah". It generated the same result as just using the "what is".

Solution: convert all questions that starts with "what's" to "what is" before sending the query to google.

Google Custom Search Engine (Business Edition)

Part 1: The Set Up
We decided to use google's custom search engine to do the search for our "similar questions" feature and for our regular search feature a while back. Fulltext search just wasn't cutting it. Who does better search than google right?

I finally got around implementing it last week.
It was a fairly painless process:

1. sign up and pay (Wayne took care of this part)
2. grab the URL from the admin panel page.
   the URL is only available to the admin or the person who paid.
   adding me as a contributor didn't work and you can't add any additional admins.
   
3. read the doc. this took me a while to find for some reason:
   http://www.google.com/coop/docs/cse/resultsxml.html
   it seems the doc for the google application is similar but NOT the same as there are more functions that can be used (like meta data filtering) in the application.
   
4. stick the URL in my existing search object, in a new function:
1. send queries to google
   
2. get xml back
3. parse and style xml
4. display results
   

It returns pretty good results, comparing to the old fulltext search.
There was a lag time of about 24-48 hours (this still needs more testing, but that's the general consensus on the interweb). New questions do not come up in the results. That's ok though because the older questions usually have more answers. Who wants to look at questions with no answers?

Everything was peachy then I get one text early next morning from Wayne telling me there are "mad bugs". uh oh.

to be continued...