Minutes from the W3C - WAP Forum workshop on Position Dependent Information Services, day 2, february 16, 2000.

Greg Fitzpatrick, "big" chief scientist, MetaMatrix. Buzzword debunking, Whole kit&kaboodle

Thanked Keith Cheverest for the loan of his computer to do the presentation - since his own had crashed.

We are talking about position advantaged information systems - not dependent.

I am more inspired by Adam Smith than Karl Marx - want too make that clear by explaining the business model.

Generalized Unversal Directory is the web. It is unstructured. But there are pockets of structure around the periphery - databases. Quality of service is also a problem (the Internet being a best-effort network).

Markets are naturally segmented by the fact that the user physically has to get from one market to another market. Usually we tend to define the real word in terms of obstacles between markets. There are a number of such separators.

Recursive effectiveness means that you recognize that limits exist, and make yourself more efficient inside them.

Paradox of broadcasting is dependent on proximity. It becomes more costly and less effective as we reach the edge of the market.

Within a given time limit, the natural time and geographical limitations are happening. Personally, I am not sure "the death of distance" is the real thing. Natural segmentation is gone - in order to remain effective, we have to create new segmentation. We do that through structure. We are partitioning markets by structured information.

Technology based on artificially maintaind obstacles is old world technology. For instance, an inherited monopoly. Old technology generally built on obstacles (creating or overcoming). New technology is based on the realization of possibilitites.

The concept of a generalized universal directory is a junction between the codified and wild information. Wild info is not codified. Maybe never should be - should remain free. Can either be minimalist, or structuralist. They can be contained in the same concept by extensibility.

Here, Greg introduced his Wha-concept, which is better described in the slides. He also told the following funny story:

The Wha's came from the cavemen who had problems speaking. They would make an empty sound when hungry - wha? Wha is the question. Very unnuanced. Eventually they developed the what when where etc. We use this in Skical, because you can create structure by the language. By the way, the XML in the slides is not wellformed, it is ppxml - Powerpoint XML.

Wha concept is based on 90 % of questions tourists ask for. Who is very important - it is important to understand where the information came from. The "who do you sue" person. Each infomediary may package in its own way - there must be a mangement plan for the object.

Often there is confusion about the selfdescription of the object, and what the object is describing. Following the iCal 256 character description (from the iCAL standard). Everything else goes into why.

Event description, describes the event in terms of intersecting what, when and where. An event is an instance of this intersection (in object terms). Using existing codes (like insurance company codes) to describe an event, such as a traffic accident.

Elucidation - creating an extension of the structure.

There are other intersections. Who with what, what with where, what with why etc. (editor note: See the slides. Greg counted to 162 the last time we looked).

Getting people to publish metadata is extremely difficult. Even if it is in their best interest. Swedish research projects have to describe themselves in Dublin Core terms. Even so, it is hard (even though they threaten with cutting off their funding).

The whow is a misspelling wow to be symmetric. Matrix for what tourist agencies ask - take a property, what is its conditional qualifier. More, but the point is that an extremely large amount of information can be written like this. Are there any restrictions? Not in the XML version. In the MIME type version, you can introduce the parameters.

SkiCAL, submitted as draft to the IETF. Open, public. Extension to the Calsch working group iCAL system (the successor to vCAL). Versions of that is in Notes, Outlook, etc. i Cal basically is intended for business meetings. Dependent on it. Took ISO 8601 (time description format), iCal added positining, place. But it is "monophonic" (you can only describe one meeting). SkiCAL people are musicians, so we tend to use music terms. There is also an XML version of the iCAL standard, which is not consensus document.

SkiCAL starts with events. Immediately that creates a conflict, because if you get the entire event schedule of Stockholm, you fill up whatever memory you or your computer has. SkiCAL is a time expender system (for people who spend time). Mime type, but also XML.

Business model is creating new world business opportunites. The old model is the Tom Sawyer infosystem. A gathering point was set up by someone who was a little ahead. Tom Sawyer would get others to paint the fence for him - these types of systems asks you to publish information on them using their own forms and data structures. There are a number of them - not just one. Come on and register - and they all use private formats. 18 calendars just in the Stockholm area.

Instead of creating a private aggreagation, we create an infostructure around the publisher, and make information machine readable. If the Tom Sawyers wanted the information they would have to get it themselves. But it would be easy. The information would be published only once.

The proximity of something is old world technology - the new world is about agents acting on the information. Who would turn over all the trusted information agents to one carrier? That will not happen. You have to create an infostructure where they can operate.

So how do you create an infostructure. How do you empower the hot dog stand and pizza parlour? We have to find a common ground of people - and who represents this. Sweden is first, since it is a club society. There are organizations that represent all of these interests (down to the most narrow ones). Tourism agencies and tourism companies were all in the close proximity of each other. They were all in the process of building a portal - but they know nothing about machine readability. But even though some of them were building portals, they understood that this was not the best way to help their constituencies.

Second wave of attack was to work for the harmonization of standards. It is not enough to have a structure, but you also need a common vocabulary. There are a numbr of different formats for this. If you go to Microsoft Biztalk, you have to go through the statistic codes. They are great to keep things separated. The tourist industr can either use a private or public schema (like those from the UN). But statistical codes is a maze of overlaps of different codes.

We are trying to bring together these groups. It is a difficult transition for the old standard organizations.

Third wave of attack, promote schemas and lists of public domain information. And do it in RDF. Cities have coordinate databases - even the government was thinking about publishing things like sales tax rates - can we do it.

There is also the "queen bee aspect": There is only one organization that is the source. There had to be a way of publish a dynamic source. When it comes to governement publishing, governements can not take responsibility since there is no public responsibility. It is important that you establish the queen bee.

Last year the city of Stockholm decided to go with the SkiCAL model. The city had five people reading faxes, taking phone calls and opening letters to recieve event information, and then typing it into the computer system. After 6 months the information providers will have to use Skical.

When it comes into the information service, information gets structured, both at the infomediary and when it is written. Anyone can get the stuff and repackage it. It is in the benefit of the city to have this and provide it for free. So how do you differentiate yourself? You put added services on top. The business is what you put on top.

In the middle ages, time was controlled by the catholic church - it became a business model of the catholic church. All time keeping was property of the church. Then Luther came, and overturned the curch. From 1540 to 1568, 500 watchmakers from all over France moved to the Calvinist city of Geneva, where there were no clockmakers before. Because information wanted to be fre.

Q: Skical and RDF?

A: I didn't want to confuse you but of course - Skical will be in RDF. We try to work modularly, not do all syntax at the same time. But we do it in RDF, XML, and Mime. Don't forget, there is more iCAL in peoples computers than XML.

Ingemar Eriksson, Swedish Travel & Tourism Council

How do you prepare a nationwide tourism industry for Internet II?

Most countries have a portal. But it takes a lot of adminstration to get all informaation. When I startedat the Swedish national tourist agency, they did not even have a database. When we were ready to take the next step, the question was what to do. We could not roll out proprietary data model. Then, we met Greg.

About travel: No matter where you go - there you are.

The NTO (National Tourist Office), has 20000 companies working with it, 10 offices. The country is full of information, so how do they take that information and put it in printed material? Sweden has a huge penetration of web usage. But we wanted to prepare for mobile Internet and time and space sensitive information without a propritary system. One way would have been jut to put a WAP gateway on the databases. But that does not solve the problem. We had to create an informatin structure.

The new technology is what the companies in the room represent. You should not work there. Nor in web databases. We had to prepare the information itself.

Today, the cycle is slow - event is born in the head of its orginator, before reaches the customer there is a process of typing, storing, etc. Slow, inaccurate, labour intensive.

With a common infostructure model, the event creator fills the information in a ski generator. If they have a database or hardcoded web does not matter, as long as they can put the information into a metafile. The harvester of the NTO goes out at night, retrieves the inormation, present on their web page. Can translate automatically to other laguages. Use positioning, GIS, can be an intermediary.

The company has to register with them before they create an event. No anonymous crawling.

The tourist company, can stop process with their web site. Newspapers, others can just retrieve the information, instead of the laborous pocess that exist today. Value added services can reuse the information. And calendar applications can put the events into a calendar. What will that mean to booking companies? The entire travel booking is seven or ten instances of time and space sentive information. Toay you give th user a bunch of papers. It's possible to put in flights, trigger reminders, see how far users are from airport, etc. You can then give a reminder so he gets there in time.

Rollout in Sweden, Stockholm Information services starts this week, Gothenburg this spring. Rest 2000. Goal all web/dbbased info are to be in skical files by end of 2000. This will create a number of new services. As a NTO, it is much better to take care of the infostructure.

Q: Are publishers resisting the use?

A: Just out of ignorance, and I talk a lot about it. When talking about machine-readability, what they gain, they buy it even though they do not understand it.

Q: Are there plans for traffic information, routing information?

A: Absolutely. When you start with an infostructure, you can add on information to this. And route planning is something that adds value.

Q: Much information already exists in other formats. Conversion plans?

A: The other information formats can be reused. A script or database object can create a skical object. You can stil use your old web pages. You do not have to get rid of anything.

Q: Do you have generators?

A: We paid MetaMatrix to build the generator. But for a database management system, you have to do it yourself.

Also note, Board of culture paid for half.

Prabuddha Biswas, Oracle

Location-aware application framework. Based on a lot of stuff you have already heard about. Not trying to adress as large problems as some of you have already talked about. Have existing technology, like Oracle spatial, Oracle portal to go, disseminated to handhelds. Want to provide to create an infrastructure. Introduce a region modeling. Internal location server.

Spatial database services, can represent any geometry, region (using polygons), put into regular table. Series coordinates that represent geometry. Indexing technology to speed up indexing of information. Support for complex geometries. Once in database, you can do analysis.

Relation queries, want to find everyting within a certain distance, nearest neighbour. Different overlaps, touch, disjoints etc. Not topology at the moment. But it is in the plans.

Portal to go, bridges devices and content. Data is there - you have to leverage that. Need a way to map data, display on devices. Create an adapter from the source format to the internal XML, transform it out.

What we want to do, is add environmental contextualization. Provide a framwork for location-aware application development.

Simplified model of geograpy, entity is client, regions of inerest that have a clear focus. once in that area, the services are important. Sometimes, the specific areas are the important. The yellow page for the area is the service, and they serve out information. Areas can be culture dependent, etc. Relationships are client-region, service-region, client-service.

Two types of regions: System regions, a hierarchical relation. Also understand that is not what people want. Cities want better granularity. Also needs to be flexible, since users can define their own regions, locations. Location dependent services can be regions or points. Region using region modelling, point is geocoded.

Internal location server (location cache), a method of acquiring the location. Activity patterns, walk to a place and ask a number of questions, walk to the next place, ask questions. Cache location for a short time, since granularity may not be so important. Going out to get the location every time is heavy. Trajectory is important for that reason.

Associate services to location at creation time. Service locations can change.

Standard is required for mapping services. Vectors are important, better than bitmap. Routing services, more value added. Geocoded services, need standarization. Service definitions - how do we find services. Want to be able to convert any input to any output.

Q: Spatial representation?

A: Work with OGIS. XML geometries can be genereated.

Q: How do you get device constraints?

A: Have a device database of hundreds of databases. You create transformers to transform from one XM dialect to another. You can publish your own transformer. (editors note: With CC/PP and XSLT you can do this entirely within the standards of the W3C).

Vincent Buller, AND Data Solutions

Started as a route planner company. Geocontent service provider. Focused on road, adress, and airline schedules.

End user, gets their services through intermediaries.

Route planner for WAP. Hoping for phones with position.

What are locations in different domains? In roads, from a technical view, they are network nodes, but from a user view road crossings. Lots of application domains that they do not cover. But there are for instance places of interest, public transport stops.

Mainly indicated as grid references, unique coding schemes (eg IATA). Free format text. Or a combination of the above.

Issues are human vs machine interpretation (humans travel to an adress, not a network node). Humans want point of intrest. Accuray (Quality of location) must be taken into account.

Each application domain has it own location definitions. Any standard that does not recognize this is doomed. A standard should therefore be flexible enough to encompass all representations.

Translations needed for a lot of stuff. Keep track of accurcy. Mobile devices may require server side location.

In the EU-funded EuroSPIN project, we have developed a central planner which accessses distriuted planner. A way of distributing calculations.

User wants to drill down - creates a pyramid of areas.

It was very easy to create valid XML location types for locations(e.g.hotels).

Q: Why coordinates referring to codes?

A: Get together with application domain or use coordinate systems which works in 90 % of the time. Airport code does not say where it is on earth.


Venky Krishnan, HP

Research started last year. We initially took a different direction: How to give value to individuals in business or corp environment? Lot of trouble, starting with the name of the project. The name was taken and used as porno site.

If wanted to access services utside the corporate environment, I ran into probems. Simple, basic notions. Obvious things, people places and things should all be first class citizens of the web. Bridge online and physical world.

Basic components, personal access devices (bridge non-IP). Beacons of method to define presence. Access points, spaces represent context.

Space seems to be overloaded. Location presence not maintained by coordinates, but by the URI pointed to by the becon. Space is defined by beacons. Becon gives you the presence, what is shown is a web page generated by the services accessed by the beacon.

Types of location: Semantic location, not based on coordinate system, but address like URL. The service determines what the meaning of a URL recieved from the user means.

Physical location, based on coordinate system. GPS, triangulation, geographic (human-readable), etc.

Services based on physical location, based on the web. Semantic services, relate. Input used, beacons. Issue of mapping between room beacon and global services. Created translation agents for services.

Created an XML-based descriptor to describe entities in the model. Define what interfaces are. Provided filters to access for global services - agents represented Yahoo. Internal mappings, no need to touch external services.

Issues: Access points. Proxy-based SSL scheme, secure access across firewalls. Different levels for different users. Not provided much in mobile. Main issue is how services are defined, accessed.

Q: By using beacons and not physical location, dont you lose info (physical relation).

A: URL allows you to code location information into it. No constraint to be added.

Q: Coordinates in URL?

A: You can use a standard URL. But it depends on the place manager.

Q: Notifications to mobile clients?

A: Yes. Software system is webbased, part of software entity is notification, registration of events.

Immo Heino, VTT

Plannning services as part of a national program on personal navigation services. Interested to see how standards support needs. Car navigation first wave. Some market already. Next wave, outdoor personal nsvigation, lots of new technologies. Perhaps, third wave, outdoor and indoor personal navigation. Local services with beacons, etc. Also wireless LAN, etc. Also positioning, European sattelite positioning, etc.

Services are coming to data networks. Map services, route services, service points, position delivery etc. Services should be available for other formats too, like PC and digital TV. Can also be auditive. Location services in the data networks. Positioning metadata and indexing. So user can find services by position. Use at home, at work, en route. Open competititon, partnershps are neeed. Need open platforms and service interfaces. Both local and global.

In detail, different map services, on the road and in towns. In cities. Different kind of service points that are available. Position delivery, emergency calls. Just in place ordering. Some kind of position exchange. Send coordinates to the mobile. Global and regional search engines.

Different kinds of methods for positioning, more in the future.

If want to see good quality of service, several issues have to be taken seriously. The service must be attractive enough, and confidential. It must be reliable. Accessible. Competition with others, and make sure of room for public-private partnerships. Add value on infostructure.

Value network, lots of different partners. Keep consumers in mind, but also data producers. Partnerships between the different players is very important to launch this market.

As for standards, we have the cellular, broadcast, wlan standards. There are a number of standardization efforts around positioning. ISO has some actvities.

Lot of work has been done, there is something to start with, but we need a common approach to create the market quickly and in a harmonized way. Very important is the query format. Metadata issues, how to find existing information from the interent, services which are located.

Spatial referencing, coordinated systems.

Content definition, what are the products, what are the maps we can have in the devices, what do you needs, what do they contain.

What is important is that the players have a shared vision on what is to be built. Must work tgether to define what is to be built.

Finish personal navigation programme. Want to test services, cooperate with other partners. Identified a number of generic technologies. Create a wider testbed for these terminals. Serve different fields.

Horizontal support needs: Regulation, usability, interfaces (at arch level).

The consortium has been going for half a year. It is a good starting point. Wider network after programme launch in April.

Kamppi project, realtime video project for user position.

Q: What is the budget of the project?

A: 30 million euro over 3 years

(the chair was by now in the Finnish timezone and called for lunch, which was an hour early. So we continued).

Patrik Pang, Singapore IDA

Vision for Singapore is "dotcomming the people, public and private sector". The city has an ATM network for entire city. Add wireless component. All Singaporeans have broadband already.

Full telecom deregulation. Building alliances, bring strategic actors together.

Apply On the Move to Singapore. Real project, IT and telecom merge. Device and location sensitivity.

The governement wants a clear migration path, see vertical applications emerging. Trying to get the feel for what is happening.

Want to design wireless LAN, stream video. Singtel is the operator partner in the consortium. Content provider, Land Transport authority. Some WAP startup companies.

Trafficscan, taxis have GPS recievers, can identify which routes are clear. Also live cameras at key junctions. Could have on handsets. Not looked at handsets yet.

Q: Position dependency in infrastructure?

A: First consortium meeting March 6. Open consortium. During the testbed, what they do should be open. After, you can create service.

Q: Who is the systems integrator?

A: CWC is responsible. Project lasts for a year. Consortium is small enough to be an integrator in itself. IDA is project manager.

Q: Will Singapore have broadband wireless access?

A: 2001 or 2002, WCDMA.

Q: WLAN over the entire city?

A: There is a company doing just that, as an alternative infrastructure.

Patrik Fältström, IETF Applications Area

There are several organizations in the world who tries to work on this. Might include everything from transferring information to the services, might be related to content negotiation. IETF looks at transfer of the location itself.

Tries to stay in the field they know about: IP, and transfer IP. That is what we try to have working groups doing, from IP layer and up to syntax. Talks about calendar initiative, skical and ical have a semantics difference. More concerned about syntax. If syntactically different, can not pass data between them.

IETF is really an open review process, who publishes RFCs. Sometimes it is tricky getting things through the IETF. Someone always finds some bug. Taking care of transferring IP over something (like IP over carrier pigeons - an actual RFC - and yes, we do publish on April 1).

Transport on top of IP. How to move something from one person to another.

Supporting applications: DNS, routing, security infrastructure. Time. Little surprised to not have heard more about here. Loction is four dimensional, time is a factor. Distributing time is difficult if things are moving around. GPS is not good enough for NTP accuracy. Can use GPS as relative clock, there it is good enough.

All work is done in a working group. Has to have a narrow charter. Can not have a WG that is going to work on position dependent services. Has to define precisely. Consensus is established on mailing lists (which also serve as documentation).

IETF knows how to transport things over IP, reuse existing functions (like security), open review process. It does not do semantics, not higher level stuff.

Explanation if you are reusing domain names, you make sure that you refernce the correct things, so that you do get changes automatically when a specification changes. Look into them so you do not forget time zone definitions, internationalization, etc.

Look at what is done: Does HTTP, mime types; W3C handles XML, HTML.

For digital signatures, there is a shared working group with the W3C.

Trying to see if it is possible to do a simpler protocol than HTTP for XML transport.

Fall 1999, IP spatial requrements for protocol - new mailing list. Started with transport. Trying to resolve problems of: Trusted location data echange between IP devices. Constraint is that all IP devices in the world have to do this Locate devices gepgraphically, map IP adresses to regions. Publish resources.

Current target, SCTP is issue. Transport area. People are worried about efficiency. HTTP is 80% of traffic, which means 40% is TCP apps. Tell him or the mailing list that they are doing the right or wrong thing. Important.

Do we have an overlap? 

My take after listening to yesterday, want to make sure they do not to toch semantics. If use time, signatures, jump in and make sure they are used right. Offer help. IETF has an open review process, menas open for reiew.

Do a lot of cooperative work between orgnizations, make sure it happens.

Rigo Wennig, W3C

What is P3P. focus what we are talking about. Privacy identified and identifiable data. Identity tied to human person. If an identity on a device on a device that is use by many, not a problem. if address is tied to single-user device, can be identified. Central term is purpouse. What is data needed for, what do we want to do with it. User agreements, battle between USA, Europe.

Users privacy concern, cookies incoded in URIs. Supress cookie does not mean anything. Make profiles and integrate by data mining. Profile of the person. Huge amount of data is also a problem. Central concept data protection, user must know what others know about him. Mobile devices, cases in the US, got location of others, to commit crimes to them. Anonymous on the net, otherwise that may happen.

Business concerns, why is this issue so emotional? People do not have the trust in the system. In Holland, free ISPs became attacked by publicity, so privacy became a brand. Data protection has become so complicated that you have to simplify it.

Governemental concerns, tapping wirelines, tracking criminals, are not adressed. Have to be adressed. National regulatory issues.

Problem in privacy various legal issues, regulations floating around. 160 countries, at least 10 - 15 different approaches. No common regulatory or self-regulatory framework. Vocabulary must express different policies for diferent needs. No way to define a global legal framework, but we can define a vocabulary to talk about the issues.

P3P, enables the user to have privacy, create privacy value. Extensible set of vocabulary to express the privacy policy. Extensible data schema, also used for ecommerce. Matches with vcard. Selection language, APPEL.

P3P is the vocabulary to talk about privacy in a structured way. Server defines policy, which is matched against preferences of user. Policy and preferences are exchanged via HTTP and HTTP-HEAD. Also possible to send in the document body as metadata (with META tag).

Privacy and position. Which issues are involved. Determine which are the issues. Which are concerns about repression forces (who has to have a interface). What is the vocabulary you need to discuss these issues. Is an extension of the existing vocabulary needed.

Q: W3C seems to have a lot of work going? When will it be implemented?

A: Hoping that it becomes implemented soon. Implement it and give us feedback.

Olivier Perrault, EBU

The EBU is a organization of national broadcasters, who mostly negotiate broadcast rights. Co-productions etc. And standardizaton. We are in the forefront in the standardization of various broadcast protocols, like DAB, DVB, etc. An example is interactive tv channels. Traffic and transport information for broadcasts. Already, there is an european protocol for RDS (Radio Data Services). Now, focusing on services with bigger bandwidth. Transport Protocol Expert group. Design inherited from others such as MPRG. Hierarchical frame structure. All information coded.

Also icludes syncronization and error detection. Possible to add data later. Bearer independent, can decode from different bearers. Just like WAP.

Only one big application, Road Traffic Message. Allows the user to get information on any event on the road network. Hierarchical, create methodologies from sets of tables. Compatible with dictionaries used by road authorities. Completely automatic transer from the road information centers to TPEG.

Other application that exist is SNI, Service and Networks information. Describes servcices in the network, what happens when user goes from one bearer to another. User go from one reciever to another.

Road location, a spatial and descriptive location. Machine readable and human readable. There will be services that just use text, descriptive can be useful. Use intersection. Location, for transport, location is related to the road. It is a vector, and it has to say which side of the road. ILOC, used only in Europe. Recommend to look at. Now datastructure is available. Other informaton soon available, Public Transport Information, Status and Travel Time information. Held workshop with eur public transport providers. Can see when a problem occurs on the networks. Broadcast time for each part of the network. Give route guidance locally.

Similarities between WAP and TPEG. Adressed independently common problems. TPEG could be bearer for WAP, TPEG could use WAP stack. Many possibilities. Protocols are completely open. TPEG will define in XML. Only application.

Martin Littlejohn, EC, DG Information society, new name for DG XII

Main task to manage 5th Framework Programme, the research programme.

3 roles: Political, helps stimulate market through support programmes, and is an actor in the market itself.

Planned a communcation to launch initiative around geographic information. Turns out this is very difficult. Working on it for a number of years, to get member states to take on it. Needs political champions. Commission resigned last year, new commissioner taken bold decision, not go forward as an independent initiative, folded into the policy on access to public sector information.

Geoinfo is needed across borders. But there is much national and little european data. Copyright restrictions. Cause for developing GIS policy on european level. Different size of areas. Many other examples of data that was harmonized Wanted to create a policy that was market oriented. Need champion, creating european metadata, realizing potential. In 5th Framework. Global level.

Policy - trouble finding champion. Interest in different ranges, often more industry interest in interoperability.

Public sector information greenpaper, issued jan 99. 10 questions to be adressed. 40 reactions were GIS related. Was interest for pan-european system. Fragmented copyright policies. Nice if it was margin-cost based. Ready by february 2000. Maybe setting up a freedom on information charter, metadata, high level policy. Needs to go through the political decision policy. High-value data is and can be copyrighted by the governement. Basic information is not as such copyrighted.

The second role of the commission is to stimulate the market. Action lines in info-mobility, available in 2000. Administred by Daniele Rizzi. Helped to create national coordinating organization in europe, EUROGI.

Commission itself is a big user of geographic information. Has itself felt how hard it is to get hold of information. Manage to set up organization of DGs that participate. Joint work inside commision.

Lot of things in are going on in the 5th framework. Link into follow-on activities, follow-up on Info2000.

IAF (Information Access and Filtering). Proposed action line for mobile and domestic platforms. Workshops next month. Other, also information visualization.

Q: if you can pull it off- great stuff. Because we can't in the US.

Jean-Francois Galliet, ERTICO

Partnership, including ITS (Intelligent Transport Systems). Promote and support ITS in europe. Help members implement and sell ITS.

Applications need accuracy of a few meters. In the future, centimetres. NEXTMAP project, assessed if comercialy viable. Need one common location reference method.

One of the first rules for european systems, locations of major roads, give attributes on location. Not easily related to maps.

Uses GATS geocodes, longditudinal and latitudinal distance from a geographical centrepoint, which is in Berlin.

All system has disadvantage of creation, maintenance, creation of tables.

ILOC, means intersection location. WGS84lat/lon, plus 3 attributes that are road descriptors (5 digits each). Each map has its own accuracy, so you can end up in different places in two maps from different information providers. New ILOC adds a category for the element reference, standardized in ISO.

For he moment, a data model and not an actual encoding.

Q: How much is done in ILOC?

A: Encoding of attributes. Hitrate is 80 %, not commercially viable. Take into account an encoding to make sure that they are talking about the same things. Introduced object type element. Embedded in GDF.

Some issues: For some applications, we want paneuropean services. Some encoding, information is needed.

In GATS, you can create a type of anonymization.

Future systems should take maps into account.

Liability. What if accuracy is not good enough - who is responsible? Needs to be sorted out.

Has to stay technology independent.

Gallileo project, European GNSS, creates new localization facilities. New messaging protocols?

Kirk Burroughs, Snaptrack

Snaptrack - wireless assisted GPS. Company was formed becasue of e911. Going after wirless assisted GPS. Believe expand more into the value-add services like WAP.

GPS, long-term governement committment to make free. Conventional GPS reciever, time precision device. Tells you where sattelites are, etc. Limitations conventional GPS. Unobstructed view of sattelites. Takes a long time to set up. Problems with continous tracing.

Wireless asisted GPS, takes the functions and breaks up between GPS reciever, location server sends back snapshot (which sattelites it can see), sends back response to server from which location is calculated. Possible to go back to handset, or out to origin server.

2-way comm needed.

Low power consumption.

With cell broadcast you can send out the sattelite view data in the cell, saves a message (GSM specific).

Q: Are you satsified using SMDDP over IS-41 or TDMA?

A: Data burst in CDMA is defined as SMS. Separating it on databurst layer makes sense to us. Not a huge problem.

We will see - like WAP Forum already created - express messaging, solves many of the problems of SMS.

Q: Is it satisfactory for emergency services?

A: You need some GPS coverage, not as much. And of course cellular coverage.

Q: Have you adressed the question of GSM blinding GPS (the GSM radio). It is an antenna problem.

A: Believe we have. Need 1-second snapshot, when not transmitting.

Louis Hecht, Open GIS consortium: Introduction to GML

Encodes position information in XML. Web friendly. Generate a map from GML in SVG using XSLT.

Its concepts are based on geography.

Notions of location, postion, extent, etc can be application dependent. Different for forrester and ambulance driver. True even in an impoverished device.

Roads, rivers, streets, buildings, etc are features. Some features have geometry values. Defined shapes, coodinates are given according to a reference system.

XML enoding. Consistent with existing standards.

Usually a backpointer from the SVG to the GML, so features can be found and identified.

Consistent with many WAP properties. Based on RDF and DTD/Schema.

Legacy data is really important. Most data is in legacy databases. Could generate routes from data sets. Foresees persitent data stores, using xlink.

The DTDs are derived from the schema definitions. Can extend using Subpropertyof, SubClassOf.

Geometry, srs DTDs can be generated using XSLT from the RDF Schema definitions.

GML critical to the semantic web.

Release 5 expected to be public on march 15.

Future, extend with temporal structures, scema definitions for datatypes.