Thursday, May 19, 2005

A Little History of Percevia

This blog provides information about a new type of search engine called Percevia. Essentially the Percevia™ search engine offers a way to search interactive media that gurantees you will always get a result.

Take a look at the example we created with Percevia for a web-based bird field guide to identify birds. The beauty of this engine is no more “zero matches” or thousands of matches. The engine uses a parametric "step-by-step" approach that helps you quickly narrow down your search with just a few clicks so you can identify any object in seconds. The Percevia site is designed to help you understand the our technology and how you can use it to power your web site. The bird search engine is just one of many examples, check these other databases. You'll find a Laptop Buyer's Guide, Movie Finder, World Fact database, even a database for identifying firearms.

Before you read this blog I would like to tell you a little history of how Percevia and the bird search engine got created. If you don't care about the past, ramblings of a technologist, or the origins of dreams, just click here to get started on the web search tutorial. However you might want to read this as it shows how frustration can be turned to inspiration, and inspiration can enable a new technology for making things work better.

The Bird, the Rain and the Pocket PC

The inspiration for Percevia came from a tiny electronic device that I carried in my pocket called a PDA, also known as a "Palm Pilot" or, if you are a Microsoft fan, a "Pocket PC". If you have not heard of these contraptions I suggest you take a look at one, its like having a laptop computer with you at all times. Anyway it was a dark rainy night. Well actually it was not that dark and not that rainy but it might as well have been--I was cold, muddy and frustrated, hiking in the hills of California, and trying the best I could to identify an elusive bird fleeting about in the branches of a redwood tree. Like most "birders" I carried a copy of Tory Petersen's "Field Guide to the Birds of North America", a classic book full of beautiful paintings of birds with paragraphs describing each in great detail.

There was only one problem. The bird would not stand still long enough to look it up.

The Trouble with Books

"That's the trouble with books", I thought. They're great when you have loads of time to sit and turn pages. But when you can only catch a momentary glance at the bird, they are the pits. Flipping through hundreds of pages while trying to remember the color of the bird you saw, its shape, bill, the kind of call it made--all fade from your memory. Not to mention how confusing it is to see a Duck, a Hawk and so on when all you want is a perching bird.

I sat down on a wet log. The PDA bulged in my back pocket. It gave me an idea--would it be possible to use my Pocket PC as a "digital field guide" to help identify the bird? Could I take advantage of a mobile device's searching features to narrow in on the bird I am trying to identify faster than I could with a book? Could I put the sounds the bird makes into the device so I could play it back in the field and see if the bird recognizes it? Is there a way I could indicate to the device just a few of the field markings and characteristics of the bird so I could match it quickly? It seemed like a good idea but I was not sure the PDA technology had enough power to perform these functions, and had my doubts about how to type information into such a tiny device. While some of these PDA's have keyboards, they are incredibly tiny, too tiny for my fat fingers.

The Dearth of Good Search Engines and the Birth of Percevia

My first thought was could it be set up as to require no typing, so you could just touch the screen, or tap it with the stylus. To do this I would need a way to indicate the bird's attributes visually, and I was not sure that could be done easily either. I started by looking at all the web-based bird search engines as well as CD ROM software. I discovered they all had a major flaw--you would enter all the field marks ahead of time, everyone you noticed, usually by clicking check boxes, then hit the search button. If you did not select enough characteristics you would get back way too many birds. If you selected too many characteristics you would frequently end up with "No Bird Meets Your Criteria" as an answer.

Frustrated, I knew there had to be a better way. Then it hit me--make the search more of a step by step process, allowing you to pick one attribute at a time: color, location, pattern, eye color, size, wing shape, etc. That way the search would narrow a list down and there would always be a valid result. One great side effect of this approach would be that I could eliminate field mark attributes after each search step that no longer applied to the bird I was looking for. That would mean each step would only present marks that where valid so if I got to the third step and it was "color" I would only see the colors of the birds I had narrowed the match down to, rather then all of them. In fact if the birds where all the same color that field mark would not even appear. Would that not be cool?

Rentacoder and Flash

I started building a mock up of my idea using a multimedia tool called Macromedia Flash. Then I posted it on the on the net in a few web sites I use for renting programmer talent, one great one called Rentacoder. My idea inspired a lot of programmers and eventually a very good one who specialized in mobile devices, Robert Levy. We started working on a real working prototype. Then we realized something--this idea did not need to be limited to birds, it could be used to identify objects in any kind of collection--be it cars, computers, coins, stamps, trees, movies, etc. Any collection that had a set of common characteristics was a candidate. At that point I decided to build more general purpose search engine, and use the bird database as a primary example.

Outsourcing Illustrations

Since all good field guides start with excellent drawings or photographs, and I personally prefer illustrations because they make it easier to emphasize the field marks, I started searching for talented artists and immediately hit a wall. I discovered that drawing birds is a very difficult job--feathers are hard to draw--a typical illustration takes well over a day to finish (I now know why Mr. Audubon spent his entire life drawing). Given the number of birds I needed was well over 900 for North America, I had to find a solution that I could afford. Then I heard about outsourcing. I don’t want to start a political war here, but in this particular case if I was restricted to artists in the USA I would not be able to create this engine, so I did not feel I was taking anyone's job away. I used a cool web site called Elance to post my idea and quickly artists from all over the world to help me: India, Argentina, Russia, and China. The best artist, and the one responsible for 50% of my drawings, came from the Ukraine.

Poor PDAs

At the time we started this work in 2001, the processing power of the PDA technology was not fast enough to handle the graphics, so we decided to first build the software on a laptop PC platform, then we would scale it down to the PDA later. This would give the PDA technology time to mature and at the same time we would learn more about the user interface. The desktop version was completed in January of 2003.

All along I had been adding new illustrations, bird calls and ornithological information to the bird database; by the time we where done it was up to 400 birds. That was another ah ha moment--we realized that the bird database was in every way superior to all similar products on the market, and would make a great resource for others to use. Our business plan is not to sell databases but the technology for building them, and so we decided to make the Birds of North American (Western Edition) database available for free educational use. But then we had another problem.

Too Big for a CD ROM

The size of the database had grown to over 100 megabytes and thus distribution would be a problem. Sure we could use CD ROMs, but we where updating it all the time so a CD ROM would quickly be out of date. The solution was to make a web-based version of the search engine that could be accessed by any browser. This solved the PDA problem as well since, with a little modification, we could use the browser in the Palm or Pocket PC to access the same web site. In January of 2004 we launched Percevia.com. From this site you can access the search engine, and the bird database, as well as several others. We made our Field Guide to Birds of North America our best example of the Percevia technology. The database of is freely accessible to anyone interested in learning about birds or just identifying them. At this time our goal is to get feedback on the product--we would like to know your impressions, what you like or dislike, improvements, mistakes, anything.

SQL Enterprise Edition

In March of 2005 we converted our Percevia database format to SQL and made a new Enterprise edition available.

Forums

To make feedback easy we set up the Percevia Forums where you can leave your feedback.I hope you find this bird search engine useful, and look forward to any comments you have.

Sincerely,

Mitchell Waite

Mitch Waite Group
160 C Donahue Street
Suite 226
Sausalito CA 94965
415 888 3233
email: mitch@percevia.com, web: http://www.percevia.com/

P.S. The Percevia bird identification software is being used by the students at Piedmont High School for their 39th Annual Bird Calling Contest, which is featured each year on David Letterman's TV show.

11 Comments:

Anonymous Anonymous said...

Just a thought (that has probably been rejected already): would a forced procedure benefit the search? You know, require Location entry before the next step can occur (with an easy-to-find skip option), and so on? Maybe with the ability to switch to all options, as set up now. I dunno. Might be a pain to set up.

8:54 PM  
Blogger Simone Whitecloud said...

Ah, yes, the age old dilemma of the bird disappearing before you find it in the book. Even with Percevia technology in your back pocket, it's still essential to train yourself to make the most of a sighting. Beginning birders are often frustrated by this exact problem. My tip for anyone who's been shared a similar experience is to program the eye to fully take in a bird while it's visible. Start at the beak and work your way over the back of the bird to the tail, asking yourself questions like 'does it have a short bill or a long one, a plain head or is it patterned, is the back striped, solid, or spotted?' etc. Then move the eye from the tail up the front of the bird, noticing it's belly and breast pattern. You can train the eye to do all this in a matter of seconds. Watch the bird until it flies away, trying to get the best description possible. Its departure marks the time to grab the field guide, be it book or pocket PC. Of course, only the PC can confirm your ID with vocalization playback.

8:56 PM  
Anonymous Anonymous said...

I think this is an awesome web site, can you tell me more about the actual technology that powers it? What language did you use to create it? How did you get it to search so fast? Is it easy to set up on a server?

9:00 PM  
Blogger Mitchell Waite said...

To answer the question about the forced procedure, which is a great one, yes I think that would be a cool feature to add.

The problem to solve is what are the most optimum order for the attributes to be presented to the searcher? For example if someone is looking for a certain bird we do not know ahead of time which are the best attributes to narrow the search fastest. In one case it might be simply habitat, e.g. for a bird that lives over the ocean, picking Ocean as an attribue will narrow it fast, however in another case it might be the head pattern which might not narrow it much.

There are some attribute-value combinations which do not narrow the search significantly, for example selecting the location will not produce a small list of matches if the state is California.

Perhaps the way to handle this is to employ some AI to analyze what searches users performed for each bird and which resolved the match the fastest. Still I am not clear on how that would be applied in a reliable way. But there probably is a way, just need to think about it for a while.

9:07 PM  
Blogger david lukas said...

I've been writing text for species accounts on the Percevia site and it's an amazing way of thinking about each species one by one. There is not only the problem of how a user might utilize search attributes, but at my end there is also the question of how to best describe the patterns, behavior, and characteristics of each species. This intensive focus is extremely instructive and it's teaching me a lot about birds that I already thought I knew. Of course the user will never get to this text unless the search engine works properly, but once they arrive at their target bird I want the information to be of the highest quality, not to mention fun to read and entertaining. Much of the stuff that's out there on the web and in bird books is rather dry, so it's a challenge to liven it up a bit wherever possible. At the very least, the species accounts should summarize the features that someone might click on during their search.

10:53 PM  
Anonymous Anonymous said...

In the size attributes, is there a better way to visually express bird size than the images used? A friend using that attribute refused to select medium because she knew what she saw was not a duck.

9:26 PM  
Blogger Mitchell Waite said...

I answered the question about size in a new blog post called The Size Question.

10:22 PM  
Anonymous Anonymous said...

One good resource for full software download and many more free software alternatives is Software4YourSuccess.com
Yes it is my site and I would love for you to drop by for a second. From there you will have free access to several of my products such as The Marketing Toolbar (which is goldmine of information on how to do things quickly and on the cheap, thus saving you time and money).
Also for webmasters I have created Webmaster Wizards, which will help you with almost every aspect of putting code on your site from things such as legal and privacy disclaimers to popunders. I have so much content and free software there I could fill up this whole blog, so check it out Thanks, Sincerely Rob Rudd

1:11 AM  
Anonymous Epson inkjet cartridges said...

You blog sure gets quite a few comments. I'm going to read thru your posts and see why you are generating so much interest. Bye. Ms. Epson inkjet cartridges

10:54 PM  
Anonymous Anonymous said...

single parent match

6:37 PM  
Anonymous Anonymous said...

government job site

4:30 PM  

Post a Comment

<< Home