Where Is Waldo? 
An Explanation of an International Location Based Service

It has always been a necessity of everyday life to be mindful of location and to navigate throughout the world. Whether it's a simple trip to the grocery around the corner or a trip around the world spanning seven continents, everyone needs to know where they are and how to get to their next destination. In the past, people have used tools such as paper maps, landmarks and compasses to navigate throughout the world. In the 21^st century the human necessity for locations and directions is using technology and human innovation to take it a step further.

Currently there are powerful mapping and locating tools that can pinpoint a person or a location within a couple of meters. People can visit a website and in a couple of clicks find a Starbucks Coffee house in London and get a detailed map of its location. These mapping tools make life easier and more efficient. Today we're going to talk about the technology behind the map that magically appeared on that website.

International Location Based Services (LBS) are the magic behind the scenes. The following sections will explore the technology behind a LBS and an International LBS and what role they play in web services. Next, we'll explore how an International LBS is created with GIS data and software, with a case study of Visa International and InfoNow Corporation. Lastly, we'll get a special view on how a International LBS should be tailored for Internet users in a global Internet community and we might even find Waldo along the way.

An International Location Based Service

A Location Based Service or System (LBS) is software that understands geographic locations (Geographic Information System Data or GIS Data) and will provide a map along with matched addresses of the input address based on calculations of latitude and longitude. A "Locator Service", as a Location Based Service or System is sometimes called, will have key components of a Mapping Server, Geocoding Server and Proximity Server.

An International Location Based System or Service will have GIS data for more than one country. It will be able to provide a web service in many languages and character sets to multi-national companies, organizations or groups of people as individual end-users.

There are two types of Internet Location Based Services. The first is an advertising supported web site that serves everyone who has access to the Internet. The second is a seamlessly integrated service into a client's Internet presence.

1. Free consumer based LBS:

Within the Internet and Wireless communities, there are companies that offer the ability to search for a location and request driving directions with a map for anyone who has access to the Internet. Advertising supports these services so they can be offered free to the general public. Proprietary web sites offer these location services for one country, or many countries. In North America, the most widely used sites are 'MapQuest' and 'MapBlast'. In Europe, a company called 'Maporamma' is widely used. In Asia 'MapPointAsia' and in Australia, 'Whereis' are the most popular sites. These websites will also allow their applications to be integrated into other company's websites as a tool to route end-users/consumers to a company's physical locations.

2. LBS's integrated into client portals.

Using a LBS as a seamlessly integrated tool inside another company's website is a popular solution for companies who want to route their consumers and partners to their physical locations as a service on their websites. These applications can provide the end-user/consumer with a map and list of the store locations closest to their present location. Such locator services may include complex business rules provided by the company specifying how locations should be presented to the end-user/consumer.

A LBS is used as a component within other Internet web services. This application can be an added feature, part of a process, or the backbone of a particular web service. The components of a LBS are used within Internet e-commerce, channel management and opportunity management to increase company organization and profitability.

E-Commerce

Currently, there are two very popular location-based web services related to e-commerce. The first involves the consumer buying a product online and having it shipped to their home or office. The second involves consumers finding a product they want online and the website offering an address or directions with a map to a retail location to make the purchase. The end-users will enter their location at some point during the process and this data will be handled by components of the LBS within the e-commerce web service software.

E-commerce software uses components of a LBS for:

1. Directing the end-user/consumer to the point of sale location.
2. Calculating which shipping option is the cheapest based on the distance of the location of the product and the end-user/consumer of the website.
3. Capturing demographics such as where end-users/consumers are, where they are purchasing, and how far they are willing to travel to see a product.
4. Boosting sales by making the purchasing process for the end-user/consumer easier and less time consuming.

Medium to large companies will often sell their products through resellers or partners. This is referred to as, "Channel Sales". Channel management software will make the relationship between the originating company, their partners/resellers and the consumer, more efficient and robust, which builds more profitable relationships. The components of a good LBS used within Channel Management software can make the relationships stronger.

Channel Management software uses components of a LBS for:

1. Providing territory management of resellers and partners for the sales processes.
2. Address verification of resellers and store locations used in sales compensation.
3. Searching for partners, locations, resellers or end-users/consumers in a territory in a channel.

Companies who use Opportunity Management software are able to quickly and automatically refer prospects to the best partner to meet their needs, send leads to the most appropriate business partners and track the progress of those leads through sales closure. A company will determine their own business rules for guiding lead referral, which enables them to match customer business requirements with partners on the basis of location and the products and services they provide.

Opportunity Management software uses components of a LBS for:

1. Routing leads to the right people by using proximity calculations of the latitude and longitude coordinates of locations. These coordinates are found through address matches within the geocoding databases of a LBS.
2. De-duping addresses to profile the customer and the leads correctly.
3. Allowing for customers and partners to plot themselves on a map to see their locations.
4. Visually find qualified partners based on their location on a map.

The most successful International LBS will have GIS data in at least 20-25 different countries of major business importance, translation of at least 25 languages including regional dialects and character sets and fast, accurate maps and match results from the input addresses. This requires very accurate GIS data, exceptional software to read the geographic data and reliable peripheral hardware architecture. The components of an International LBS are essential to achieve profitable e-commerce, tight relationships with channel partners and resellers and most importantly, sales prospects that become revenue.

The creation of an International Location Based Service is no easy feat. The two most important components of and LBS are GIS Data (Geographic Information Systems Data) and software. Acquiring the data and setting up your software for processing the GIS data and serving the Internet are the key steps in setting up a LBS. It is essential to have the most accurate GIS data set to work with the most technologically advanced software. Otherwise the LBS will be just another web service, undistinguishable from all the others.

A. Contacting Vendors

When working with international GIS data vendors, it is important to remember that there are different business practices and methods for capturing and creating GIS data. It is also essential to remember that there are different standards for quality and release cycles for a data set.

A LBS needs to be fast. The maps should be drawn on the fly, in real time. Though Raster Data (GIS data that is pixilated) can be fast, this data can only create static maps where labeling cannot be done in real time. Vector data is the best type of GIS data for a LBS as it can be used for dynamically drawn maps that can be tailored to each clients requirements for their portals. When contacting GIS data vendors, ask for information on Vector data sets.

Researching Vendors

If someone is going after a data set for a particular country or region, it is best to do some research about companies who produce the data or find companies to work with who know the best data source. A good place to start is the Esri International reseller's website. Often the Esri reseller for a region or country will be a consultant who knows the vendors or producers of a GIS data set for their area. Keep in mind that there can be more than one good data set for a country or region, but not all GIS vendors and consultants are aware of them. It is best to do some extensive research and contact as many companies as possible. Be aware of offerings, quality, pricing and licensing. It is equally important to contact the vendors by telephone on their local time. Using e-mail will produce slower response.

Data vendors are sometimes Data Creators. Data creators are Independent companies who have created a GIS database from the very beginning. Retailers of GIS data can also be a public branch of the country's' government mapping agency (example Ordinance Survey in the United Kingdom) or a third party who is repackaging, selling and licensing the data from the data creators (example MapInfo Corporation). It 's best to work with a company who is the original creator, and currently maintains the data sets and has sole ownership of the product. When problems arise with a data set, working directly with the technical department of the data creator will give a faster turnaround to fixing the problems. It is also easier to get lower prices on the data sets with better licensing privileges when working with the creator of the data sets.

Multi-Country Data Providers

When looking for a multi-country data provider, there are a few who stand out. Some examples are TeleAtlas, Navigation Technologies and GDT. While these companies spend a considerable amount of resources on creating GIS databases from the very beginning, they also contract with or purchase other companies to add additional data for enhancement. It is wise to compare these companies side by side, country by country, region by region. The main differentiator will be the amount of the population that is covered by the GIS data. This is referred to as 'Coverage'. Whereas one company might have the best coverage in France, they might also have the worst coverage in Italy. It is necessary to look for the best overall coverage rates, the best update cycle to the data and the most flexibility in the contract and licensing of the data.

When contacting vendors it is essential to inquire about the quality of their coverage. Before asking about their coverage rates, you need to know what your company's thresholds are for the quality of the data sets. The best International LBS's offer street matching capabilities for a minimum of 50 percent of the population. Meaning, in each country, the input street address will be matched in the street geocoding databases 50 percent of the time. Therefore, the GIS street data set you purchase should also cover a minimum of 50 percent of the population. For postal code and city GIS data, a good threshold is 80 percent of the population or more covered by the data. Establishing thresholds for the data quality before you contact vendors will help eliminate lesser quality vendors.

Quick guidelines for contacting GIS data vendors:

1. Look on the Internet for companies as well as ask for recommendations/more information about all the companies that sell data from consulting groups.
2. Emailing will produce slower results than calling the companies. Call the vendors until all questions are answered.
3. For just one region or country, it is best to go with the original creator of the data for ease in the forward working relationships.
4. If you need a multi country provider, go with the larger companies such as TeleAtlas instead of many smaller companies.
5. Ask for coverage statistics of the population by the GIS data set in question. Make sure thresholds have been set for the needed data set before contacting the vendors.

After researching and contacting the right vendors, you should ask for a sample of the data and test it with your system. Ask the prospective business partner to send you three sets of data:

1. Major City - population of over 500,000 if possible, preferably in a large city you know are familiar with.
2. Medium City - population of between 50,000 to 80,000
3. Small City - population of 7,000 or less.

1. See how well a company covers all levels of the population as claimed at small, medium and large scales.
2. Look for the attribute systems of the data to work with your software.
3. Check if the data type will work with the external software for the Internet and internal processing software.

C. Contracting and Licensing

Having chosen the vendor that best fits your needs, the next step can be the most consuming and tedious. Negotiating a licensing contract for Internet use can take months. Keeping with the laws of economics, consumers want the lowest price possible and the vendors want the highest price possible. There are server-based licenses and transaction-based licenses, but rarely do consumers get the lowest price and the vendors get the highest price.

Server based licenses variants include:

1. Pay a large sum up front for the original purchase of the data, then pay a yearly maintenance fee for updates to the data in addition to royalty payments.
2. Pay for power units of the servers that are using the data sets in addition to a yearly maintenance fee and possibly more royalty payments.

1. Pay an initial cost for the data, maintenance fee for updates, then receive a number of free Internet hits to the data for each month and for however many hits the company produces over the free allotment, pay a fee.
2. Pay for all transactions on the servers that use the data.

Case Study of a Data Vendor and LBS Provider Relationship

When InfoNow Corporation searched for the best GIS data source in Denmark for their International LBS, they choose to contract and license with a small data creator, Kraks Forlag AS. Kraks provided InfoNow Corporation with an excellent data set that included already geocoded street segments that covered over 80 percent of the population of Denmark. After the initial purchase of the data, each year InfoNow Corporation pays maintenance fees for updates of the data and royalty payments. Kraks Forlag AS are quick to work with InfoNow Corporation on a problem with the GIS data if they arise. They are able to edit their data immediately, as they are the data creators. This is the most ideal working relationship.

Once the GIS data is purchased and licensed, the creation or purchase of the software to read the data will be the next step in creating a International LBS. Only after the GIS data is licensed, can the software be put in place to use the data and ultimately find Waldo.

Every Internet company who intend to launch a LBS should choose the best software and architecture solution for their internal and external systems. Whether the LBS solution includes proprietary software or licensing software from an outside vendor, companies need to be mindful of how their GIS data sets will be used within their internal and external systems.

There are a number of companies in the market place that are willing to sell customized software solutions from geocoding software to real time mapping software. If a company chooses to use these external software packages, the company will lose some flexibility that is associated with proprietary software. Writing software for geocoding and mapping Internet solutions is difficult, but it can save the company time and money in the long run. Money will be saved with proprietary software when royalties are not paid to software providers for the each client who uses the LBS. Companies with proprietary software will avoid the cost of buying and contracting for new software upgrades as technology and the GIS data sets get better, as their in-house employees will be able write the upgrades themselves. Lastly, writing proprietary software over purchasing software will save money when instead of having to wait for the next release of the vendor software to fix a problem, the company housing proprietary software can make the fix's right away to keep their clients happy. The best option for creating a reliable, inexpensive and well-maintained system is to have internal resources work with a consulting company specializing in this type of software to create a proprietary system that can be easily upgraded over time with better technology and more accurate GIS data sets.

Once a company has chosen the software for geocoding addresses and serving maps in real time on the production servers (the cluster of servers behind a firewall that receive requests from websites on the Internet). The next step will be to choose software for the preprocessing of the data. The goal of preprocessing is to make the overall system faster for the end-user. Each database does not need every attribute, every node, and every vertex. These extras will inevitably make the mapping and geocoding databases slower as it searches through the large data sets.

Processing the Mapping Databases

Operating System

The creation of the mapping can be a lengthy process if a company has purchased heavily attributed data sets. Assuming the vendor's original data comes in a file format such as Esri's shape file format, your data sets can potentially arrive rather large. When working with Gigabytes of data, you need to ensure that there is a very stable environment to handle processing that can take longer than anticipated. UNIX systems have proven to be the most stable processing environment. Purchasing processing software that runs in a UNIX environment will save time and help keep the integrity of your data. Other operating systems have a tendency to crash and lose data. The most powerful data processing software available is Esri's ArcInfo for UNIX. The power of the command line and stability of UNIX makes almost anything possible.

Data Types and Databases

If the data comes in a format different from the desired format, there are software converters that can be purchased. Examples include MapInfo's MapInfo Professional 7.0 software or FME. Once the mapping data is in the right format, it can be loaded into the processing software, where an automated system of scripts will save time and effort as well as human error when each country is created. Your automated scripts should drop attribute fields that are not needed, add fields that are needed with the appropriate data populated. Part of the processing of the mapping data should include connecting road segments of the same attributes to save space in the database. This will make the drawing of maps faster in real time. There are numerous types of databases that will store the mapping data sets. When deciding upon the production systems software architecture, choosing the right database can be the most crucial part, as the need is to house very large sets of data and give the mapping software the ability to search through the data to draw the map quickly. Today's Internet user expects quick loads of web pages and few are willing to wait for a map that takes more than 10 seconds to draw and load. Esri's Spatial Database Engine (SDE or Arc SDE) running on an Oracle database in a UNIX environment is a good mapping software database, though there is a broad range of database types available.

Testing for Production

It is crucial to check the integrity of the mapping and geocoding databases before the data is loaded onto the production servers. There should be a very extensive regression testing process in a testing environment that mirrors the production system. In the testing environment, it is essential to follow a regimen that will be used in each mapping database update. This should include timing the maps drawn; checking the polygons to ensure that they are being drawn properly and that the software is drawing unclosed polygons properly. Check to see that the labeling of the mapping software (or sometimes called the labeling engine) is working with the new street segments to place the name on the street correctly. Another helpful tool in testing newly updated or created mapping databases is to view the data set in its entirety through a software application such as Esri's ArcView or Arc GIS. This will show if there are missing regions, landmarks, or any other holes in the data set.

Loading new databases onto Production Servers

Once the mapping data sets have been tested for speed, geographical accuracy, and are clearly understandable to the end-user, they are ready for the production systems servers. A good rule of thumb when loading new databases is to take the mapping servers offline for loading data sets in the evening, and only on a Tuesday, Wednesday or Thursday. Switching a mapping server offline or to partial down time when the system is under a heavy load of Internet requests, could potentially harm your systems' uptime. The heaviest load times are typically Monday and Tuesday morning. Loading new data on Tuesdays, Wednesdays or Thursdays will help avoid a disastrous situation for clients of a LBS. For example, if a problem with the system occurs on Friday afternoon because new data was posted earlier that day, employees will not be in the office to fix the problem until Monday, leaving potentially three days of errors on client WebPages. A heavy workload and a database update at the same time can lead to a system failure and downtime on the mapping servers. Whether you have one dedicated mapping server or 100, following these rules will help ensure a smooth running system that your clients will appreciate.
 
 

Processing the Geocoding Databases

A company can choose to create a geocoding database that fits into their proprietary geocoder or purchase the geocoding database from their geocoding software provider. The easiest method is to utilize the geocoding databases from the geocoding software vendor if a contract is in place. But if a company chooses to house a proprietary geocoder, it will be necessary to create the geocoding data sets themselves. This can be a very complicated and tedious task that involves a lot of processing hours on dedicated servers and a very extensive testing process before rolling the databases to the production servers. However, creating geocoding databases will give the LBS provider the ability to edit, update, or add to the databases when clients call in from their home locations to report an error in the data. (This tends to happen very frequently.)

Operating System

As with the creation of the mapping data sets, the best environment for processing large geocoding data sets is the UNIX environment. Esri's ArcInfo for UNIX works the best, aside from creating your own processing tools (InfoNow Corporation has found that creating proprietary processing tools with Java is a very fast and reliable alternative to using software from outside vendors). An International LBS is comprised of many data sets varying in size from 5 Megabytes to over 20 Gigabytes, depending on the country. When creating or updating a country's databases, the best processing method should incorporate as much automation as possible to avoid corruption. Each company that tackles the task of creating their own geocoding databases should first implement a processing process that is void of human error in a very stable environment. UNIX scripts are recommended for processing of large geocoding databases.

Already Geocoded Data Sets

Before creating geocoding databases, check to see if the data vendor sells data sets with street segments already geocoded. Already geocoded street segments will have the postal codes and building numbers attached. If so, check the coverage of the geocoded segments. If the coverage rate is less than 80 percent of the population covered by geocoded street segments, then a secondary data set to supplement the already geocoded database will have to be used.

Geocoding Software and the Databases

The software on the production servers for the geocoding matching will dictate what format the databases need to be in and how well the geocoding databases will be able to match the input addresses. Is the geocoding software technologically advanced enough to match each floor in a multi-level building, or can the software only geocode to a city center? These software matching capabilities will dictate how accurate the database needs to be which will then dictate how large your geocoding databases will be and how they need to be processed. The format for the geocoding databases is also crucial. Inevitably, there will be a need to make small corrections that clients and end-users will pick up. It is best to have an editable geocoding database so, for example, if there is a latitude and longitude that is wrong for a postal code or city center point, the databases can updated without having to regenerate the entire geocoding database from the original GIS data.

Testing for Production

Once new or updated geocoding databases have been created and put into an editable format, they must be tested tirelessly. The best method for testing the geocoding databases is to batch geocode a large set of addresses with the old databases and then batch geocode the same addresses with the new databases to compare the results. In comparing results, be sure to look at the addresses where it was geocoded to a better or worse match, to be able to pass the new database as good quality. Putting these databases into a test environment that mirrors the production servers for a minimum of one week where the geocoding databases are hit repeatedly in regions all over the country will ensure safe and stable geocoding databases on the production servers. Follow a standard procedure every time you create a new or updated an existing geocoding database.

Summary of Software for a Location Based Service

Though the databases for the mapping and geocoding components of an International LBS is the most essential part and often the most expensive, the software used for the processing of the data and the serving of the data to the end-user can be considered almost equally important. Trust the software purchased and put the software through numerous tests. Most of all, automate as much as possible when creating the geocoding databases. An International LBS that runs in a stable environment that draws fast maps and returns quick, highly accurate geocoded matches to the end-user is not unattainable. It takes reliable software, excellent GIS data, and powerful hardware.

VISA International provides the world's most widely used form of "plastic payment". VISA International Service Association financial institution is the payment leader of credit cards worldwide by providing access to cash at ATM's or purchasing power with credit cards around the world, reaching millions of consumers. Visa's use of a International Location Based Service provided by InfoNow Corporation is an excellent example of the most comprehensive International LBS with GIS data and software in the world.

Visa has over of 400,000 locations worldwide. Consumers who use Visa's web services can find these location that are closest to their location in 133 countries, with localized GIS solutions in 22 countries, 8 languages and 6 dialects. For the year 2001, Visa's web-based International LBS had 2.8 billion searches with 2.1 billion geocoded matches of the end-users address and drew 1.4 billion maps. In addition to a web-based LBS, Visa's consumers can also search for ATM locations on their cellular phone enabled with WAP (Wireless Application Protocol), or calling an IVR (Integrated Voice Response) locator or by using their PDA (Personal Digital Assistant) enabled with WAP.

InfoNow Corporation also provides Visa International with web services that use components of their International LBS. The 'Profiler' is a web service that allows Visa employees to edit address data in real time (over the Internet) when locations around the world change. Once a change has been made, the databases are dynamically updated and the end-user sees the changes immediately. The 'Visa Configurator' is a web service that allows member banks of Visa International to build their own customized LBS web site, which will upload their data in real time. The Profiler and the Configurator were both deployed being multi-lingual web services with multiple character sets.

Since 1996, Visa International and InfoNow Corporation have been providing credit card users around the world with an International Location Based Service that is unparalleled. Visa and InfoNow have a partnership that always provides the most technologically advanced LBS. Any Waldo using Visa's International LBS can find an ATM location from wherever they are in the world.
 
 

When in Rome, do as the Romans do! The truth of this old saying can be applied to an International LBS in that each LBS in a particular country needs to be tailored to that county.

There are three elements of an International LBS that should be tailored to each country. Each country should have their own national display of the maps, standardization of the languages and the proximity with business rules. Often, American companies try to tackle a foreign market with an American feel. This attitude or strategy will be harmful to business overseas. An International LBS should acknowledge every countries individuality, all their languages and spellings as well as some ethnic traditions and principles. Mastering the following three areas will give any International LBS the edge over the competition and help to make all end-users of the system feel represented.

1. Colors and Features of a Map

1. Color Schemes

In the world of Cartography, one color scheme does not fit for every type of map. Nor does one color scheme fit for every country in the world. Each country with their own individual traditions and unique history has different color schemes that will appeal more or less to their citizens. In the United States of America, it could be assumed that Americans like map colors to be soothing, plain and simple to read. Depending on the country, a nation's citizens can prefer the map to be more colorful or less, have more detail or less, and use different features to mark the map than the standards used by American Cartographers. It is important to do some research into the country's own mapping color schemes that are the norm. This will ensure that clients in each country like the service or product they are paying for.
2. Map Features

The points, lines and polygons of a map that will be drawn as Points of Interest, Roads and Routes, and Water, Parks and Buildings will be referred to as 'Map Features'. Residents of a country have their own ranking of importance of landmarks such as churches, golf courses, and schools, as well as the level of importance assigned to certain roads and routes. For instance, one country may consider their churches to be major landmarks and everyone knows the location in relation to himself or herself whereas another country may not put any significance on their churches geographically. Another example is one country recognizes secondary roads with major importance, another country will only recognize the difference between a local road and a national highway. These sentiments need to be recognized on a map for each country. One blanket set of requirements for all countries in an International LBS will not be as useful to the end-user, nor will the user want to use a map that they cannot navigate. Research into how a country prefers to navigate maps by looking at Atlas's from that country will give an indication of how the maps should be drawn online.

1. Standardization's of Language and Spelling

1. Language

Often residents of a country will speak more languages than the official language of a country. This is often seen throughout Europe and parts of Asia. This needs to be accounted for in each country's LBS. When the geocoding database is based on data that is created with only the country's official language, you need to add some language standardizations to the geocoding software. These standardizations will handle different spellings of the same city in different languages by accepting the users spelling, then changing it to match the geocoding databases. Depending on the country, the standardization segment can become rather lengthy. An example would be in the country of Luxembourg, the residents can speak German, French, Dutch or Luxemburgisch. The geocoding software for Luxembourg will have standardizations for these languages for spellings of names and abbreviations of popular streets and city names.
2. Spelling

Misspellings should be handled in a LBS. Often end-users will spell a city name or a street name by how it sounds to them and that will not always be the correct spelling. The geocoding software should handle the phonetic spelling of a name and return to the user the correct spelling as well as a geocoded match to the input address.
3. Business Rules and Proximity

The business rules of a LBS are client specific rules for how the end-users address should be matched to the GIS data in the geocoding database, or how their product lines are displayed. For example, the end-user should only be shown the match results for only the product they inquire about within 10 miles from their location. This is calculated in the proximity component of the LBS. These business rules will sometimes vary between each country in a International LBS. The reason is; based on a countries demographics and traditions, certain business rules will or will not apply. Countries with good transportation or a high number of automobile ownership will allow for end-users to travel far distances for a product. The proximity calculation can be a powerful tool when a client/company decides upon their business rules based on their own clientele in a particular country. Being able to offer this capability will only add to the strength of an International Location Based Service.

Have you found Waldo Yet? Finding him will not be easy, but it will be profitable, highly accurate and representative of his country. Whether building a International Location Based Service from the very beginning, or implementing upgrades and additional features to a existing service, remember…..Where is Waldo?