Agentlink Netagents Technical Forum

Survey on Agent Technology Deployment

Initial Findings

25th June, 2004

http://x-opennet.org/netagents/



This document presents preliminary results from a survey conducted as part of Networked Agents 2004, an AgentLink III Technical Forum Group. The focus of the survey is on the theory and practice of agent system deployment, in order to better understand current practices, best practices, and community requirements for agent system deployment.

Responses were gathered in May, June and July 2004 from 30 organizations engaged in agent technology research, development and deployment using a web based submission form. These preliminary results:

  • Provide raw data from actual responses to each of the questions.

  • Have been anonymized where appropriate to shield the identity of respondents

  • Do not include analysis / interpretation of the data..

The results are based on a small sample of responses (30 responses - characteristics of respondents are given in Section A) hence in some areas results are likely to be strictly unrepresentative of general trends, and in all cases interpretation of data should be carried out bearing in mind the limited sample size. More details and subsequent analysis of the raw responses and/or data from further responses will be made available by the Agentlink Netagents web site at http://x-opennet.org/netagents/

The data was collected for academic research purposes only. No use is made of the information for marketing, advertising or other commercial purposes. The information collected will only be published in a manner which does not enable an individual person or organization to be identified. Moreover, no information collected will be given or sold to third parties.



A: Organisational Information

All questions in the following sections are optional but would be very useful for us to help interpret results. Survey responses will never be released associated with identifying data without express permission from the person completing the survey.


Number of Responses

  • 30


Type of Organisation


19

(63%)

University/College

7

(23%)

Company

4

(13%)

Research Institute

0

(0%)

other:

0

(0%)

no answer

Country Organizations are based in [30 responses]


Number of Employees [28 responses]


Approximate Number of people involved in Agent Research [29 responses]


Approximate Number of people in Artificial Intelligence Research [29 responses]


Approximate Number of people in Internet / Web Services / IT technologies [29 responses]


Member of Agentlink III

23

(77%)

Yes

4

(13%)

No

3

(10%)

no answer


Previously a Member of Agentlink I or II

20

(67%)

Yes


7

(23%)

No

3

(10%)

no answer


Approximate Number of national projects you participate in that use agent technology [29 responses] &

Approximate Number of international (EU and other) projects using agent technology [27 responses]


B: Organizations Role/Activities

This section includes a number of questions regarding the type use use your organization puts agent technology to.



Do you see yourselves as mainly:

14

(47%)

Providers of Agent technology

1

(3%)

Consumers of Agent technology

15

(50%)

Equally both

0

(0%)

no answer


What type of systems have you developed (tick those which apply):

0

(0%)

None (we are new to agent technology)

2

(7%)

None (our work is mainly theoretical)

21

(70%)

Early prototypes – alpha (internal)

20

(67%)

More advanced prototypes – beta (shared with third parties)

13

(43%)

Trial systems (real users)

9

(30%)

Production systems (real users and deployment)

1

(3%)

Other: (production systems in EU projects)


Where have you deployed Agent Systems? (tick those which apply):

25

(83%)

Inside your own organization (prototypes – experiments)

4

(13%)

Inside your own organization (for your own organizations use in a real situation)

18

(60%)

With an external customer / part at another site (prototypes & experiments)

10

(33%)

With an external customer / part at another site (for real use)

20

(67%)

With several partners / spanning multiple sites (prototypes & experiments)

3

(10%)

With several partners / spanning multiple sites (for real use)

1

(3%)

Other: (real use in large-scale trials but not commercial deployment)


Name some of the Agent technologies / systems and applications you have worked on / developed or deployed (Acronym/Name/URL where possible) [29 responses]: Responses included...


C: Technology Providers

This section contains questions mainly addressed to technology providers (respondents which consider themselves both providers and consumers are welcome respond to both this and the next section).



Our contributions / solutions are mainly in the following areas (tick several as appropriate):

5

(17%)

Philosophical foundations

13

(43%)

Agent based control architectures

9

(30%)

Rational action and agency and decision making

9

(30%)

Theoretical Models / Paradigms / Algorithms

12

(40%)

Coordination techniques

12

(40%)

Communication techniques

17

(57%)

Methodologies and Modeling

17

(57%)

Runtime/Platform tools

7

(23%)

Reasoning/Decision making tools

13

(43%)

Simulation

20

(67%)

Combination of Agents with other technologies

10

(33%)

Analysis, specification, design and verification techniques for agent systems

4

(13%)

Other: (Logic languages for Internet agents, distributed and dynamic resource allocation and optimization & user modeling)


Areas in the field you believe are covered well (solutions are READY FOR USE by non-specialists in significant development/deployment) - tick several as appropriate:

6

(20%)

Philosophical foundations

6

(20%)

Agent based control architectures

1

(3%)

Rational action and agency and decision making

4

(13%)

Theoretical Models / Paradigms / Algorithms

3

(10%)

Coordination techniques

9

(30%)

Communication techniques

8

(27%)

Methodologies and Modeling

21

(70%)

Runtime/Platform tools

4

(13%)

Reasoning/Decision making tools

8

(27%)

Simulation

4

(13%)

Combination of Agents with other technologies

3

(10%)

Analysis, specification, design and verification techniques for agent systems

2

(7%)

Other: (Logic languages for Internet agents, distributed and dynamic resource allocation and optimization)


Name technologies from areas which you are already using for significant deployment / production environments - i.e. what is ready for use NOW (Acronyms/Names/URL if possible) [21 responses]: The most common answers were:

  • JADE-LEAP,

  • JADE / FIPA Agent platform implementations (for several major demonstrators but not production).

  • None

  • TAP1 and LARS , J2EE, web services, XML, JADE, UML, mobile IP, IP sec, SOAP, Radius, GSM/GPRS, W-LAN (802.11), Bluetooth, UMTS, etc.

  • Recommender agents with trust http://arlab.udg.es

  • Artimis

  • SWI-prolog: http://www.swi-prolog.org/ agent-libraries

  • Netmarche: automatic negotiation EASI: Environment as Active Support of Interaction

  • JACK Agent Platform


Name some areas where you see strong potential for application but technology is not yet ready [23 responses]: Selected answers were:

  • e-Government, e-Health and e-Business: with multi-organisational scenarios, with rich communication semantics possible. Also those with high regulatory content - which must be managed in a transparent way.

  • Gaming, tourism, tele-learning, medicine, system management

  • Java Agent Services

  • Implementation of agents in mobile devices

  • All complex real application areas that require flexibility and that are mission-critical at the same time

  • Cooperation and Coordination within organizational networks.

  • Self-organizing logistic networks

  • Knowledge Mining and Management, Text Understanding,

  • Simulation on the grid Advanced web services UMTS services

  • semantic web services semantic integration

  • Technology is not the issue: (industrial) success stories are

  • Decision support systems, complex environment monitoring and control, games, cognitive modeling

  • Home computing Health care Logistics

  • entertainment, robotics

  • ubiquitous computing

  • Reasoning, planning Social organization of agents Smart & communicating objects

  • Grid

  • Agent/MAS fault-tolerance

  • Component Based Agent Development

  • Human-agent interaction Semantic Web

  • Although methodologies and modeling, and runtime/platforms tools are already quite mature, deployment in open systems where security, scalability, and management are included/supported is still a "challenge".


Our technologies / solutions to date have been primarily for the following application areas (tick several as appropriate):

2

(7%)

e-Environment

10

(33%)

e-Health

3

(10%)

e-Government

6

(20%)

e-Learning

1

(3%)

e-Science

2

(7%)

e-Utilities

11

(37%)

e-Business & Supply Chain Management

10

(33%)

Logistics

10

(33%)

Industrial control & scheduling, embedded systems

2

(7%)

Multimedia / Digital Libraries

2

(7%)

Games and Entertainment

13

(43%)

Internet and World-Wide-Web agents

15

(50%)

Information / Knowledge Management

6

(20%)

Expert assistants & human-computer interfaces

9

(30%)

Mobile User Services

3

(10%)

Intelligent home & office

11

(37%)

Simulation

3

(10%)

Virtual Organizations

2

(7%)

Banking and Insurance

2

(7%)

Education

9

(30%)

Tourism

0

(0%)

Construction and Engineering

0

(0%)

Fashion / Textiles

1

(3%)

Security / Crime Detection

1

(3%)

Food / Agriculture

4

(13%)

Telecoms network management and control

2

(7%)

Bioinformatics

5

(17%)

Robotics

4

(13%)

Other: (crisis management, meteorology, e-Justice (agents and ICT in Justice))


What are the major barriers (if any) that you perceive to applying agent technologies in the areas that you would like to? [24 responses]


  • Immaturity of basic tools Inability to inter operate current tools with mainstream business systems Lack of robustness / scalability / professional grade software Lack of willingness/understanding by bodies which might apply the technology as to what it brings (automation is a scary concept).

  • The maturity of the agent system has to improve. Our system uses new technologies like web-services, but they have still a bad performance and a large footprint. They are not yet suitable for multi-device systems that include mobile phones or PDAs. This has to improve.

  • To find out application needs that cannot be solved but by agent technologies.

  • Social ignorance of agent technology Lack of clear standard in wireless communication

  • Lack of mature and supported agent modeling and methodologies as well as lack of CASE tools usable by the majority of software engineers that are not necessarily agent experts.

  • Complex behavior of MAS themselves is hard to implement/debug with weak tool support

  • Missing CASE tools Missing Tools for rapid prototyping Testing environments / Simulation tools

  • Low level of business publics awareness of what agent software and its advantage is

  • Education of software engineers Too much theoretical work but lack of real case studies implementations Need for more CASE tool support

  • industry level agent communication language

  • Scalability Security

  • Poor theoretical foundations, lack of solid and ready-to-use models, maybe all the research in this field is narrowly-focused, avoiding important foundational problems in order to develop (too) quickly specific solutions. Lack of strategic vision.

  • Cultural -- the typical problem of any innovation Problems of communication and understanding between academy and industry

  • private information protection

  • the use of agent technologies often seems (or is) be complex

  • lack of standard methodology to agentify a problem difficulty to verify and control MAS the users are reluctant to use agent-based systems

  • understanding of how to design and build agent systems critical mass of available agents

  • Security

  • maturity of concepts, processes, and tools

  • wide-spread deployment of inter operable platforms, agents should be a ubiquitous as web servers and html

  • - acceptance by "public" (i.e., agents must be trustworthy before they are delegated to perform a task) - security and web of trust (actually previous point) - scalability of agent systems (both the multi-agent application and the supporting platform) - m


In what sectors do you expect agent technology will bring the greatest benefits? [23 responses]

  • Those with a highly distributed, multi-organization nature, in which systems are only partially controlled. Those with high regulatory content (norms, rules, laws). Where delegation of control is required but mission critical decisions are being taken.

  • Any domain that needs a highly distributed architecture can benefit from agent technology. One popular example are massively multiplayer online games, which are basically an application distributed on thousands or ten thousands of devices. A appropriate multi-agent system as a platform could make the implementation very simple and fast.

  • Integration of heterogeneous & legacy systems

  • Logistics Knowledge management Internet and information agents

  • Telecoms, logistics and e-financing.

  • Supply Networks Service Industry e-Maintenance & Diagnosis

  • Logistics and Supply Chain Management, Operational Scheduling, Business Network Modeling

  • Telecommunications: Information services and UMTS Economics

  • Service Coordination (logistics, scheduling, etc.) Tourism Heath care

  • All sectors where information logistics is done "by hand", i.e. people play interfaces between existing systems.

  • Knowledge management: a variety of different agents competing and cooperating for searching/selecting/evaluating information

  • Social impact of technology

  • Web

  • e-commerce e-transport e-democracy information retrieval on Internet cooperative simulation

  • requires too much thought for me to answer :-)

  • Collaborative applications

  • High Level Design and Development of Business Applications.

  • All sectors with inherent distribution and a certain grade of complexity in information exchange, e.g. Supply chain management Health care Distributed scheduling in common Personal assistance agents

  • decision support in various domains


In what sectors do you expect agent technology will see the most take up? [16 responses]

  • Eventually those above, however there will likely be a wave of agents in simpler applications such as tourism and entertainment - which emphasise interaction, user modelling rather than agent-agent communication.

  • The gaming industry plays very often the role of an "early adopter" in new technologies. I think that a multi-device gaming platform for multiplayer online games based on a multi-agent system with a powerful AI engine will be the ideal product for the gaming industry. Here we also have a working business model and companies can make real money with their applications and services. Investments in future technology are therefore feasible.

  • Integration of heterogeneous & legacy systems

  • Recommender systems

  • Most probably in quite different sectors where the key requirements is the ability to deal with dynamic and changing environments and constraints.

  • Domotics

  • Banking and Insurance

  • B2B

  • Man-Machine interfaces

  • e-commerce

  • Internet applications in general

  • Planning, Scheduling, Coordination.

  • business-to-business operations distributed control systems

  • (also see answers to previous question)


D: Technology Consumers

These questions are aimed mainly at technology consumers (respondents which consider themselves both providers and consumers may respond to both this and the previous section).



What type of application are you applying agent technology to (or are planning to) - tick those appropriate:

6

(20%)

Single agent systems

16

(53%)

Multi-agent systems

5

(17%)

Multi-organization, single agent systems (1 agent per actor, many actors)

8

(27%)

Multi-organization, multi-agent systems (multiple agents per actor, many actors)

7

(23%)

Agent Populations in simulations

0

(0%)

Other: (none)


Which elements of the problem are you seeking to address using agent technology (tick those which apply):

8

(27%)

Interfaces (Human – System)

7

(23%)

Dynamics

9

(30%)

Semantics

11

(37%)

Negotiation

13

(43%)

Coordination

6

(20%)

Automation

8

(27%)

Large scale / distributed deployment

9

(30%)

Complex system (modelling / dynamics)

3

(10%)

Scalability on single platforms / organisations

4

(13%)

Scalability across many (many users/servers at remote sites)

6

(20%)

Robustness

2

(7%)

Security

11

(37%)

Intelligence

7

(23%)

Simulation

0

(0%)

Other: (none)


What are the main areas in which agent technology is importance in your applications? (tick 2-3 which are MOST relevant):

2

(7%)

Philosophical foundations

3

(10%)

Agent based control architectures

2

(7%)

Rational action and agency and decision making

4

(13%)

Theoretical Models / Paradigms / Algorithms

8

(27%)

Coordination techniques

11

(37%)

Communication techniques

6

(20%)

Methodologies and Modeling

8

(27%)

Runtime/Platform tools

3

(10%)

Reasoning/Decision making tools

4

(13%)

Simulation

6

(20%)

Combination of Agents with other technologies

1

(3%)

Analysis, specification, design and verification techniques for agent systems

0

(0%)

Other: (none)


What are the applications areas you are working in? (tick one or more as appropriate):

2

(7%)

e-Environment

8

(27%)

e-Health

1

(3%)

e-Government

2

(7%)

e-Learning

0

(0%)

e-Science

0

(0%)

e-Utilities

6

(20%)

e-Business & Supply Chain Management

5

(17%)

Logistics

4

(13%)

Industrial control & scheduling, embedded systems

1

(3%)

Multimedia / Digital Libraries

2

(7%)

Games and Entertainment

8

(27%)

Internet and World-Wide-Web agents

9

(30%)

Information / Knowledge Management

5

(17%)

Expert assistants & human-computer interfaces

4

(13%)

Mobile User Services

1

(3%)

Intelligent home & office

8

(27%)

Simulation

4

(13%)

Virtual Organizations

1

(3%)

Banking and Insurance

2

(7%)

Education

5

(17%)

Tourism

0

(0%)

Construction and Engineering

0

(0%)

Fashion / Textiles

2

(7%)

Security / Crime Detection

1

(3%)

Food / Agriculture

2

(7%)

Telecoms network management and control

1

(3%)

Bioinformatics

0

(0%)

Robotics

2

(7%)

other:(crisis management, meteorology, e-Justice (agents and ICT in Justice))


What are the major barriers (if any) that you perceive to applying agent technologies in the areas that you would like to? [12 responses]

  • (see previous section)

  • Social ignorance of agent technology Lack of clear standard in wireless communication

  • Complex behavior of MAS themselves is hard to implement/debug with weak tool support

  • Low level of business public's awareness of what agent software and its advantage is

  • Reliability and usability of standards (i.e. FIPA).

  • industry level agent communication language

  • Scalability Security

  • the users are reluctant to use agent-based systems

  • lack of broadly acknowledged industrial standards missing support of key software or IT companies for agent technologies various poor agent oriented software engineering approaches no clearly defined field of problems where agent technology is a "must have"

  • development methodology, tools ("AgentUML")


E: All Respondents: Technology Readiness/Requirements

Putting yourself in the mindset of choosing THIRD PARTY technologies for use in a NEW small-medium scale R&D activity for you organization and or limited demonstration for a customer (example 1-2 people for 12 months or more) please indicate below what your priorities are in making such choices.

Note – exclude your OWN in-house software (since usage motivations are different) – focus on third party software you may reuse. Also take “software” to mean packages specific to the project – exclude generic technologies such as web servers, office suites, etc.



In most projects the percentage (%) of non- in-house software we expect to use (by number and importance of packages) is approximately:

1

(3%)

0-10%


3

(10%)

10-20%

4

(13%)

20-30%

5

(17%)

30-40%

7

(23%)

40-50%

2

(7%)

50-60%

3

(10%)

60-70%

1

(3%)

80-90%

1

(3%)

90-100%

3

(10%)

no answer


The packages we are most likely to use include (Acronym/Name/URL where possible) [25 responses]:


When picking technologies which levels of maturity (of the third party software) would you consider:

1

(3%)

Alpha grade - basic functionality / no testing

2

(7%)

Alpha+ grade – some basic functionality / some testing

15

(50%)

Beta grade – basic functionality sound / somewhat tested

19

(63%)

Beta+ grade – Beta but used in several applications by third parties

18

(60%)

Production grade

1

(3%)

Other: (dedicated/limited to using open source software (stable though))


When picking technologies, which of the of the following license types are you likely to prefer (select 2-3):

21

(70%)

GPL-like open source

20

(67%)

LPGL/BSD-like open source

13

(43%)

Academic free only (commercial use at cost)

3

(10%)

Closed source / Binary (free)

1

(3%)

Closed source / Binary (once-off cost)

0

(0%)

Closed source / Binary (yearly fee)

2

(7%)

Licenses are not an issue in decision making

2

(7%)

Other: (may depend on the customers/partners, free source)


What are your main motivations behind this license preference? [23 responses]

  • Subsequent ease of access to our own code / systems (for others) - they do need a costly tool, contribution to the public good, ability to reconfigure/change the tool we are using if necessary ease of sharing our results lower cost

  • The value of a technology used for development is not driven by a license type.

  • I simply support GPL

  • Economic motivations: little funding available for software expenses

  • flexibility

  • Insight to source code to be able to fix bugs independent of support

  • Especially in technology transfer projects costs are a major issue.

  • Our own software is likely to be under the LGPL, since we are an academic research facility

  • availability and reliability

  • Such kind of software is continuously evolving and improving. No cost. No maintenance dependencies.

  • -depends on client (e.g. a lot of clients do not understand "open source") -cots -availability -maturity

  • commitment to open source, reliability

  • Mostly legal motivations -- this is going to be so complex in the next years, that we do not intend to be involved in any legal problems with every product / solution our projects will produce.

  • The free use of them and the possibility to have exploitation

  • access code freely available

  • cost ability to distribute what we develop

  • Possibilities of extension, flexibility, cost, re-utilization.

  • GPL/LGPL idea.

  • - general availability (size/openness of user community) - access to source code - possibility to be used in commercial projects

  • Our own product is intended to be LGPL, thats why we avoid integrating GPL-Products. We want to ensure that commercial use is still possible, although we have decided to be open source.

  • low cost often a broad community some kind of guarantee for further improvements and development early availability of on-the-edge technologies

  • Academic freedom to publish, use and re-use code, etc.


The level of budget we would allocate for software licenses in such a project on average would be approximately [19 responses]:







  • Of the 19 responses: 6 people responded 0, one 250 Euros, 3 with 500 Euros, 5 with 1000 Euros, and one each with 5,000, 10,000, 15,000 and 30,000.

  • The average amount over all those that responded was: €3,500 Euros

  • The average amount over those that responded excluding the two highest and two lowest (zero) outliers was: €1,450 Euros


When picking software packages, the level of support we would find acceptable when deciding on a package (tick one or more):

3

(10%)

no support

11

(37%)

volunteer / email support 1-2 people at the maker of the software

22

(73%)

volunteer / email support / small-medium user community

2

(7%)

professional support / small-medium company (with support contract)

2

(7%)

professional support / medium-large company (with support contract)

7

(23%)

support is not an issue in decision making

1

(3%)

Other: (expertise of a customer/partner)


The budgets for third party SUPPORT per research project on average is around [19 responses]:







  • Of the respondents, the vast majority responded zero.

  • Of those that responded with a non zero figure, the average is not particular meaningful, however comes out at approximately: 2100 Euros.


When choosing software packages, the following are the most important priority (above the others) - tick one or more but not all:

20

(67%)

Open Source License

0

(0%)

Closed Source License

17

(57%)

User community

20

(67%)

Maturity

8

(27%)

Support

1

(3%)

Developed by a major IT vendor

7

(23%)

Personal contact to the team making the software

3

(10%)

Software is produced by a consortium partner

14

(47%)

Standards compliance

1

(3%)

Other: (free source)


Some of the third party systems we have experience of using include (Acronym/Name/URL/functionality where possible) [21 responses]: responses included the following:




Some of the systems we would consider using in the future include (Name/Acronym/URL where possible) [11 responses]:


Some of the non-Agent systems we are considering applying to agent purposes in the future include (Acronym/Name/URL where possible) [10 responses]:


Our overall view on the availability of third party tools for use in Agent technology is:

0

(0%)

Very poor - lack of viable tools is a major barrier for us

9

(30%)

Poor - there are some viable tools but they do not cover our needs

15

(50%)

Reasonable - there are viable tools available to cover our needs but there is little choice

3

(10%)

Good - our needs are meet by a good selection of tools

0

(0%)

Very good - our needs are met by excellent tools

1

(3%)

Other: (don't know)

2

(7%)

no answer


Do you have comments on the reasons for your answer to the previous question? [9 responses]



  • Agent tools unfortunately often cover the lower levels of functionality - which are increasingly well covered by mainstream software. There are few good "libraries" which can easily be used for higher level functions (e.g. reasoning) and easily be embedded in multiple organisations. Certainly tools such as JADE have been useful - however there seems to be a significant way to go before they can A) be considered production grade and B) offer considerable added value over mainstream (e.g. web services) tools.

  • None

  • JADE platform in general is fine and suits our needs. What is missing are tools to support fast development of smaller sized applications. For global applications a infrastructure allowing to develop and deploy applications developed by different independent sw providers is missing.(How do I use a service developed by some people in another country)

  • What is completely lacking currently is tool support for Agent CASE

  • Basically we have a lack of knowledge about tools for multi-agent developing support tools

  • The demands of our clients are mostly related to standard IT issues, such as -reliability -maintenance -persistence -coupling to non agent systems -coordination by non agent systems Most open source /research toolkits only cover (a limited number of) agent issues.

  • It is just the time of agent research, Nothing strange.

  • For areas like agent development platforms there are some good tools, and some choice, but nothing that covers all aspects.

  • most of the agent development kits are research projects that evaluate or focus one or few agent related issues, but there is no toolkit, which has rich functionality and a distinct development approach


F: All Respondents: Related technologies

Agent technologies are often seen as a structuring mechanism or counterpart to technologies in other fields. In particular for deployment, agents are generally combined with other technologies.



Where do you see strong AGENT tools / solutions models which you feel you could apply? (tick 2-3 which MOST apply):

2

(7%)

Philosophical foundations

8

(27%)

Agent based control architectures

2

(7%)

Rational action and agency and decision making

5

(17%)

Theoretical Models / Paradigms / Algorithms

11

(37%)

Coordination techniques

14

(47%)

Communication techniques

7

(23%)

Methodologies and Modeling

12

(40%)

Runtime/Platform tools

3

(10%)

Reasoning/Decision making tools

12

(40%)

Simulation

5

(17%)

Combination of Agents with other technologies

3

(10%)

Analysis, specification, design and verification techniques for agent systems

0

(0%)

Other: (none)


Where do you see strong NON-AGENT tools / solutions models which you feel you could apply (and hence agents are less necessary)? (tick 2-3 which are most appropriate):

3

(10%)

Philosophical foundations

2

(7%)

Agent based control architectures

3

(10%)

Rational action and agency and decision making

5

(17%)

Theoretical Models / Paradigms / Algorithms

3

(10%)

Coordination techniques

8

(27%)

Communication techniques

5

(17%)

Methodologies and Modeling

7

(23%)

Runtime/Platform tools

4

(13%)

Reasoning/Decision making tools

4

(13%)

Simulation

2

(7%)

Combination of Agents with other technologies

5

(17%)

Analysis, specification, design and verification techniques for agent systems

0

(0%)

Other: (none)


Agent related technologies which we have worked with in combination with agents include:

18

(60%)

Web services

4

(13%)

REST web technologies

4

(13%)

Multimedia technologies

14

(47%)

Semantic Web

7

(23%)

GRID technologies

10

(33%)

P2P technologies

2

(7%)

ebXML

1

(3%)

rosettaNET

1

(3%)

other e-Business technologies

13

(43%)

AI planning systems

12

(40%)

OMG UML

3

(10%)

OMG Corba

5

(17%)

Other Middleware

15

(50%)

Databases

1

(3%)

Other: (back-office / legacy software)


In the next 6-12 months we are likely to investigate the following in combination with agent technology:

17

(57%)

Web services

0

(0%)

REST web technologies

3

(10%)

Multimedia technologies

16

(53%)

Semantic Web

12

(40%)

GRID technologies

7

(23%)

P2P technologies

3

(10%)

ebXML

0

(0%)

rosettaNET

3

(10%)

other e-Business technologies

8

(27%)

AI planning systems

2

(7%)

Other: (administrative systems, inference engines)


Some of the technologies/systems/tools we have worked with (or are planning to in the next 6-12 months) to achieve this include (Name / Acronym / URL where possible) [14 responses]:


Our main motivations for combining other technologies with our agent approaches include (tick 2-3 of the most relevant):

0

(0%)

We are not working on such a combination

22

(73%)

Provides necessary functionality

6

(20%)

Provides more stability

8

(27%)

Provides visibility to new communities (non-agent)

4

(13%)

Our customers / partners require it

13

(43%)

Our Application requires it

1

(3%)

Other: (integration with existing systems)


For these technology combinations, we intend to:

8

(27%)

Develop your own systems from scratch

17

(57%)

Extend existing agent systems with agent functionality

16

(53%)

Extend existing non-agent systems with agent functionality

8

(27%)

Use third party (non-agent) tools – applying an agent methodology

1

(3%)

Use third party tools which already combine agent and non-agent technology

1

(3%)

Other: (pragmatic approach using all of these)


The types of tools / systems we prefer to implement in-house include:

13

(43%)

Runtime environments

11

(37%)

Management and monitoring tools

6

(20%)

Debugging tools

9

(30%)

Integrated Development Environments

8

(27%)

Parsers and Language tools

10

(33%)

Ontology / Service Description modeling tools

13

(43%)

Reasoning tools

1

(3%)

Other: (agent frameworks)


The types of tools we would prefer to re-use from third parties (tick those appropriate):

13

(43%)

Runtime environments

11

(37%)

Management and monitoring tools

16

(53%)

Debugging tools

17

(57%)

Integrated Development Environments

17

(57%)

Parsers and Language tools

11

(37%)

Ontology / Service Description modeling tools

9

(30%)

Reasoning tools

0

(0%)

Other: (none)


G: Technology Providers: Current Deployments

This section is about deployment of agents in a wider context that a single laboratory – specifically deploying them in open environments where they may discover and interoperate with other agents.



Currently our agent technologies are applied (experimentally and / or production) - tick those appropriate:

1

(3%)

nowhere (or we do not provide any)

9

(30%)

In single organizations, single machine trusted environments (our own organization).

6

(20%)

In single organization, single machine trusted environments (installed at third parties).

13

(43%)

In single organization, multiple machine trusted environments (our own organization).

5

(17%)

In single organization, multiple trusted environments (third party organizations).

7

(23%)

In multi-organization, single machine per organization trusted environments (our own and third parties).

5

(17%)

In multi-organization, multiple machines per organization trusted environments (our own and third parties)

1

(3%)

In multi-organization, single machine per organization open environments (our own and third parties – who are do not necessarily all know each other).

4

(13%)

In multi-organization, multiple machines per organization open environments (our own and third parties – who are do not necessarily all know each other)

1

(3%)

Other: (on-line on the web)


We believe our technologies/application have the potential for deployment in the following environments (tick those which apply):

1

(3%)

nowhere (or we do not provide any)

10

(33%)

In single organizations, single machine trusted environments (our own organization).

10

(33%)

In single organization, single machine trusted environments (installed at third parties).

10

(33%)

In single organization, multiple machine trusted environments (our own organization).

13

(43%)

In single organization, multiple trusted environments (third party organizations).

11

(37%)

In multi-organization, single machine per organization trusted environments (our own and third parties).

12

(40%)

In multi-organization, multiple machines per organization trusted environments (our own and third parties)

7

(23%)

In multi-organization, single machine per organization open environments (our own and third parties – who are do not necessarily all know each other).

14

(47%)

In multi-organization, multiple machines per organization open environments (our own and third parties – who are do not necessarily all know each other)

0

(0%)

Other: (on-line on the web)


The typical current deployment size (in number of organizations) of our technology / applications is:

5

(17%)

1

7

(23%)

2-3

7

(23%)

5+

3

(10%)

20+

1

(3%)

100+

0

(0%)

1000+

0

(0%)

>10,000

7

(23%)

no answer


The target deployment size (in number of organizations which we aim to achieve) of our technology / applications is:

0

(0%)

1

1

(3%)

2-3

6

(20%)

5+