Conversational AI and Healthcare Chatbots

Used Eliza.  Still need to see better studies of chatbots in complete and varying contexts.  And comparison to having a professional available.  And also compared to professional in the room where they can watch reactions to that they say. 

Conversational AI Making Headway in Powerful Healthcare Chatbots   By John P. Desmond, AI Trends Editor  

Conversational AI has come a long way since ELIZA, which was intended by its creator in 1964 to be a parody of the responses of a psychotherapist to his patient, as a demonstration that communication between a human and a machine could only be superficial.  

What surprised Joseph Weizenbaum of the MIT AI lab was that many people, including his secretary, assigned human-like feelings to the computer program. It is acknowledged as the original chatbot.   

Pranay Jain, cofounder and CEO, Enterprise Bot

In the 50 years since then, chatbots have evolved first to engage users in dialogues for customer service in many fields, and now to dialogues on personal medication information. “With the advent of cognitive intelligence, chatbots were given a facelift. They were able to analyze context, process intent, and formulate adequate responses,” stated Pranay Jain, cofounder and CEO of Enterprise Bot, in a contribution to AI Trends. The Switzerland-based company was founded five years ago.   

Still, chatbots incorporating AI today are challenged to successfully process technical commands, to understand human intent, to exhibit conversational intelligence and understand different languages, accents and dialects.   

Today, “The ability to understand the subtle nuances of human tonalities, speech patterns, and mimic human empathy in the form of texts and voices is what makes a chatbot truly successful across industries and verticals,” Jain stated.   

Chatbots in healthcare had been perceived as high risk, with healthcare professionals skeptical that patients would provide confidential medical information to a virtual assistant. “Today, chatbots are being designed and deployed to perform preliminary pathology and aid healthcare professionals,” Jain stated, noting that chatbots now gather initial personal information and then ask about symptoms.   .... ' 

Reservoir Computing from Ohio State

OK, mostly new to me, sounds useful. Following up. 

Scientists Develop the Next Generation of Reservoir Computing

By Ohio State University, September 22, 2021

Researchers have found a way to make reservoir computing, a machine learning approach to processing that requires small training data sets and minimal computing resources, work between 33 and a million times faster than a traditional reservoir computer.

In one test of next-generation reservoir computing, researchers solved a complex computing problem in less than a second on a desktop computer, a task that would otherwise require a supercomputer, says Daniel Gauthier, professor of physics at Ohio State University. "We can perform very complex information processing tasks in a fraction of the time using much less computer resources compared to what reservoir computing can currently do," he says.

The system is described in "Next Generation Reservoir Computing," published in the journal Nature Communications. 

Using an artificial neural network, scientists fed data on a dynamical network into a "reservoir" of randomly connected artificial neurons in a network. The network produces useful output that the scientists can interpret and feed back into the network, building a more and more accurate forecast of how the system will evolve in the future.

From Ohio State University  

Designing a Quantum Computer

 in ACM TECHNEWS  Implications? 

Quantum Computer Helps Design Better Quantum Computer  By New Scientist, September 22, 2021

Researchers at the University of Science and Technology of China in Shanghai used a quantum computer to design a quantum bit (qubit) that could power better quantum systems.

The resulting plasonium qubit is smaller, less noisy, and can retain its state longer than the team's current qubit model.  The researchers used a variational quantum eigensolver algorithm to simulate particle behavior in quantum circuits and smooth out negative properties while developing advantageous features, without needing to build many physical prototypes.

Each plasonium qubit is only 240 micrometers long, or 40% the size of a transmon qubit, which will allow processor miniaturization.

The plasonium qubit's strong anharmonicity also means the additional states the qubits can reach are varied and less likely to be found accidentally, reducing the potential for computational errors. ... 

Research at the University of Science and Technology of China in Shanghai has led to the invention of a new type of qubit which is physically smaller, less noisy, and able to retain its state longer than the current qubit design.

From New Scientist  

Beware Food and Agriculture

 In ZDNet: 

FBI warns of ransomware attacks targeting food and agriculture sector  ... 

In addition to the May attack on JBS, the FBI listed dozens of ransomware incidents that have taken place over the last six months targeting the food sector.

By Jonathan Greig    ... 

Podcast: Future of the Office

 How well did working at home really accomplish things? 

What’s the Future of the Office?

Wharton’s Peter Cappelli talks about his new book, ‘Future of the Office: Work from Home, Remote Work, and the Hard Choices We All Face.’

Audio Player ... of podcast

Wharton management professor Peter Cappelli is the author of the new book, The Future of the Office: Work from Home, Remote Work, and the Hard Choices We All Face. Cappelli, who has for decades studied the forces shaping and changing the workplace, says the choices employees and employers must make about the future of work could be among the most important they face.

Brett LoGiurato, senior editor at Wharton School Press, sat down with Cappelli to talk about his new book. They discussed work during the COVID-19 pandemic, the complications with return-to-office hybrid models, and how employees and employers can make the best choices about what to do.

An edited transcript of the conversation follows. 

Brett LoGiurato: Could you share your overall message about what you believe is at stake for the future of the office?

Peter Cappelli: I don’t think it’s going to surprise many people to get the sense of how big an issue this is, about whether we go back to the office or not. If you think about the value of commercial real estate, what happens if we don’t need offices and all the supporting services and the little businesses and restaurants that support offices? And commuting? All those sorts of things matter. In addition to whether this might be better for employees, one of the things we know is that not everybody agrees that they want to work from home. There is the issue of whether it’s actually going to work for the employers, and that’s not completely clear.

Part of the message of the book is that we don’t know how well things worked during the pandemic’s work-from-home phase. A lot of organizations said that things were fine. A lot of employees said they got their own work done. But closer examination is suggesting that maybe it wasn’t quite so great and things didn’t work quite as well, and more to the point, there were a lot of things that were unique about the pandemic that are not going to carry over afterward.

For example, most people felt a special effort to pull together and try to get things done [because] we were keeping businesses together and keeping our jobs together. Is that going to continue afterward? Post-pandemic is unlikely to look much like what happened during the pandemic. We know a fair bit about that situation because we’ve studied it. We’ve studied telework for quite a while. That is regular businesses operating more or less as they did, with some people working at home and some people working in the office. The results there were not as nice as you might expect. People working remotely don’t do as well, and their careers don’t do as well, either.    ... ' 

Unlimited Digital Sensing

Not following this yet, but intriguing suggestion ...

Unlimited Digital Sensing Unleashed for Imaging, Audio, Driverless Cars

Imperial College London (U.K.), Caroline Brogan, September 17, 2021

A new method devised by researchers at the U.K.'s Imperial College London (ICL) and Germany's Technical University of Munich could support unlimited digital sensing. The team used modulo-sampling analog-to-digital converters (ADCs) to determine whether sensors could process more information via a different type of voltage; the researchers built a prototype with an algorithm that causes the ADC to switch to modulo voltage once a stimulus limit is reached, and folds these signals into smaller ones. Potential applications include enhanced imaging, audio perception, and improved driverless-car cameras. Said ICL's Ayush Bhandari, "By combing new algorithms and new hardware, we have fixed a common problem—one that could mean our digital sensors perceive what humans can, and beyond."

3D Printing Gas Sensors

New sensor capability. Small, cheap, monitoring for the smart home and elsewhere.

Scientists 3D-Print Microscopic Gas Sensors

Trinity College Dublin (Ireland), September 15, 2021

Scientists at Ireland’s Trinity College Dublin, the AMBER (Advanced Materials and BioEngineering Research) Science Foundation Ireland (SFI) Research Center for Advanced Materials and BioEngineering Research, and GE Research in New York have three-dimensionally (3D)-printed microscopic gas sensors that can be monitored in real time. Trinity's Colm Delaney said the team produced the color-changing sensor materials via direct laser-writing, by focusing a laser to produce tiny 3D structures from laboratory-developed soft polymers. Trinity's Larisa Florea said, "The ability to print such an optically responsive material has profound potential for their incorporation into connected, low-cost sensing devices for homes, or into wearable devices for monitoring analytes."

Spot Robot Dog Gets an Upgrade

Considerable look from Venturebeat on the Boston Dynamics Robot Spot:

Boston Dynamics has just released the latest update for its famous quadruped robot Spot, giving it better capability to make inspections and collect data without the need for human intervention.

Called Spot Release 3.0, the new update adds “flexible autonomy and repeatable data capture, making Spot the data collection solution you need to make inspection rounds safer and more efficient.”   ... 

Though it has been strongly criticized for anything that looks like it might support police.  Not that we want to be any safer.   And Boston Dynamics is suggesting it will not be configured for home use any time soon.  So it remains as they say an expensive niche device for industrial inspection, tracking and protection.  

Now we are Cooking with Lasers

 A long time amateur chef

Now We're Cooking With Lasers   By Columbia Engineering, September 20, 2021

Researchers at Columbia Engineering have digitized food creation and cooking processes, using 3D printing technology to tailor food shape and texture and lasers of various wavelengths to cook it.

"Precision Cooking for Printed Foods via Multiwavelength Lasers,"    Science of Food, explores various modalities of cooking. The researchers printed chicken samples as a test bed and assessed a range of parameters. They found that blue lasers (445 nm) are best for penetrative cooking, and infrared lasers (980 nm and 10.6 μm) best for browning. Laser-cooked meat shrinks 50% less and retains double the moisture as meat cooked in conventional ovens.

"Our two blind taste-testers preferred laser-cooked meat to the conventionally cooked samples, which shows promise for this burgeoning technology," says Jonathan Blutinger, a Ph.D. student in the Creative Machines Lab at Columbia University.

 "What we still don't have is what we call 'Food CAD,' sort of the Photoshop of food," says Professor Hod Lipson, lead author of the study. "We need a high level software that enables people who are not programmers or software developers to design the foods they want. And then we need a place where people can share digital recipes, like we share music."

From Columbia Engineering   View Full Article

City Wide Quantum Data Network

 China and Quantum Data Networks

City-Wide Quantum Data Network in China Is Largest Ever Built  By New Scientist, September 20, 2021

China has been running a city-wide quantum communications network in Hefei, the largest quantum network demonstration built to date, for almost three years.

Designed by scientists at the University of Science and Technology of China, the network's commercial fiber-optic hardware connects 40 computers at government buildings, banks, and universities clustered into three subnetworks, each separated by about 15 kilometers (9 miles). Smaller subnetworks and switches support routes between different users as needed, while three trusted relays streamline the architecture.

The researchers claim the network can be linked to other similar frameworks through long-distance quantum connections and satellite relays, clearing a path toward a global quantum network.

In New Scientist

Next Generation Brain-Computer Systems

Has been kicked around for a long time, how ready is it? 

Researchers Take Step Toward Next-Generation Brain-Computer Interface System

Brown University, August 12, 2021

A team of scientists from Brown University, Baylor University, the University of California, San Diego, and wireless technology company Qualcomm has moved a step toward a future brain-computer interface system that uses a network of ultra-small sensors to record and stimulate brain activity. These silicon "neurograins" independently capture and transmit neural signals wirelessly to a central hub for coordination and processing; the hub is a thumbprint-sized patch attached to the scalp, which also wirelessly powers the tiny chips. The team used 48 neurograins to record signals in a rodent characteristic of spontaneous brain activity, and to stimulate activity through tiny electrical pulses. Brown's Arto Nurmikko said, "Our hope is that we can ultimately develop a system that provides new scientific insights into the brain and new therapies that can help people affected by devastating injuries."

Removing Location Tracking

Thwarting cellphone tracking.

Is Your Mobile Provider Tracking Your Location? This Technology Could Stop It.

By USC Viterbi School of Engineering, August 16, 2021A new system devised by researchers at the University of Southern California Viterbi School of Engineering (USC Viterbi) and Princeton University can thwart the tracking of cellphone users by network operators while maintaining seamless connectivity.

The Pretty Good Phone Privacy software architecture anonymizes personal identifiers sent to cell towers, effectively severing phone connectivity from authentication and billing without altering network hardware.

The system transmits an anonymous, cryptographically signed token in place of a personally identifiable signal to the tower, using a mobile virtual network operator like Cricket or Boost as a substitute or intermediary.

USC Viterbi's Barath Raghavan said, "Now the identity in a specific location is separated from the fact that there is a phone at that location."

The system also ensures that location-based services still function normally.

From USC Viterbi School of Engineering

Finding Unexpected Training Data?

Can we find data that will ultimately lead to novel results.   Like the idea of fitting it to possible applications. In a  primitive way we looked at means to learn from huge amounts of historical existing data we had.    We note the 'at scale' issue. 

Unveiling Unexpected Training Data in Internet Video   By Tali Dekel, Noah Snavely

Communications of the ACM, August 2021, Vol. 64 No. 8, Pages 69-79  10.1145/3431283 

One of the most important components of training machine-learning models is data. The amount of training data, how clean it is, its diversity, how well it reflects the real world—all can have a dramatic effect on the performance of a trained model. Hence, collecting new datasets and finding reliable sources of supervision have become imperative for advancing the state of the art in many computer vision and graphics tasks, which have become highly dependent on machine learning. However, collecting such data at scale remains a fundamental challenge.

In this article, we focus on an intriguing source of training data—online videos. The Web hosts a large volume of video content, spanning an enormous space of real-world visual and auditory signals. The existence of such abundant data suggests the following question: How can we imbue machines with visual knowledge by directly observing the world through raw video? There are a number of challenges faced in exploring this question. ... '   (Much more detail at link) 

Pitfalls of Binary Decisions

Good thoughts, but simple is also good. 

Five ways to avoid the pitfalls of binary decisions

Before you decide, check how the question is framed to ensure you have all the information you need and have considered all your options.

by Eric J. McNulty

Deciding is easy: true or false? The first challenge in answering this question is that it’s impossible to know without more information. Which decisions, with what stakes, and on what timeline—these are just a few of the contextual factors most of us would want to consider before answering. The second challenge is that it’s probably not a true-or-false proposition.  ... '

Ajit Joakar on Digital Twins

Nicely placed background on analytical method and design based on physical models.

The significance of Digital Twins in design and simulation     By Ajit Jaokar

Open in Linkedin:

Background

I have been tracking Digital Twins closely over the last year – based on the recent work from Paul Clarke who is a mentor to our course and to me personally for the AI and Edge at the University of Oxford   ...   AjitJoakar

The idea of Digital Twins itself is not new – but we see that this technology will have significant impact over the next few years – especially as a technique to unify AI and IoT.

 Specifically, in my course I am interested in the role of Digital Twins in an engineering context for enhancing Model-based Design and simulation where AI will drive IoT but also extend to AR (Augmented Reality) and VR (Virtual Reality)

We are launching a new course on Digital Twins: Enhancing Model-based Design with AR, VR, and MR which focusses on the engineering aspects of digital twins. If you are interested in it, please contact me.

In this post, I outline some ideas for Digital twins in design and simulation in an engineering context. This is a complex topic, so will revisit in subsequent posts.

Overview

The idea of "Digital Twins" originated with NASA. Digital Twins were then adopted into the manufacturing industry as a conceptual version of the PLM (Product Lifecycle Management). However, the core idea behind Digital twins remains the same, i.e., a virtual model that incorporates all the necessary information about a physical ecosystem to solve a particular problem.

Engineering systems have always used abstraction techniques to model complex problems. 

But the Digital Twin takes this idea further by allowing you to model a problem and simulate it. A variety of machine learning and deep learning techniques (collectively referred to as artificial intelligence AI) play a part in the simulation aspects of the digital twin. AI helps to simulate scenarios via the Digital twin but also to make autonomous decisions. Further, we could also use Augmented reality (AR), Virtual Reality (VR), and other strategies for modelling engineering problems.

Collectively, the techniques described above are referred to as 'Model-based design.' Model-based design help engineers and scientists to design and implement complex dynamic systems using a set of virtual (digital) modelling technologies. As a result, you can iterate your design through fast, repeatable tests. In addition, you can automate the end-to-end lifecycle of your project by connecting virtual replicas of physical components in a digital space.  Once the system is modelled as a twin, various existing and new engineering problems can be modelled and simulated, such as predictive maintenance, anomaly detection, etc.

Terminology

We consider the following terminology:

Model-based design: A set of technologies and techniques that help engineers and scientists to design and implement complex, dynamic, end-to-end systems using a set of virtual (digital) modelling technologies. Collectively, these technologies can simulate and model physical objects and processes in multiple industries.

A digital twin is a digital representation that functions as a shadow/twin of a physical object or process. Digital twins are designed to model and simulate a process to understand it and predict its behaviour. Digital twin originates from engineering and is related to model-based systems engineering (MBSE) and surrogate modelling. The usage of digital twins is now more mainstream in software development, especially for IoT. Digital twins can be combined with AR and VR to model physical processes.  

Virtual Reality (VR) creates an immersive experience through VR devices like headsets and simulates a three-dimensional world. VR is used in instructional content and educational material for field workers, oil and gas, defence, aviation, etc.

Augmented Reality (AR) overlays digital information on a physical world. Typically, AR uses conventional devices like mobile phones. Pokemon GO is an example of AR usage.

Mixed Reality (MR) allows the manipulation of both physical and digital objects in an immersive world. Hololens is an example of mixed reality.  ... '

Biases in AI Systems

Excellent piece, broadly useful beyond AI applications.  

May 12, 2021, Volume 19, issue 2

Download PDF version of this article PDF  

Biases in AI Systems     

A survey for practitioners

Ramya Srinivasan and Ajay Chander   in CACM

A child wearing sunglasses is labeled as a "failure, loser, nonstarter, unsuccessful person." This is just one of the many systemic biases exposed by ImageNet Roulette, an art project that applies labels to user-submitted photos by sourcing its identification system from the original ImageNet database.7 ImageNet, which has been one of the instrumental datasets for advancing AI, has deleted more than half a million images from its "person" category since this instance was reported in late 2019.23 Earlier in 2019, researchers showed how Facebook's ad-serving algorithm for deciding who is shown a given ad exhibits discrimination based on race, gender, and religion of users.1 There have been reports of commercial facial-recognition software (notably Amazon's Rekognition, among others) being biased against darker-skinned women.6,22

These examples provide a glimpse into a rapidly growing body of work that is exposing the bias associated with AI systems, but biased algorithmic systems are not a new phenomenon. As just one example, in 1988 the UK Commission for Racial Equality found a British medical school guilty of discrimination because the algorithm used to shortlist interview candidates was biased against women and applicants with non-European names.17

With the rapid adoption of AI across a variety of sectors, including in areas such as justice and health care, technologists and policy makers have raised concerns about the lack of accountability and bias associated with AI-based decisions. From AI researchers and software engineers to product leaders and consumers, a variety of stakeholders are involved in the AI pipeline. The necessary expertise around AI, datasets, and the policy and rights landscape that collectively helps uncover bias is not uniformly available among these stakeholders. As a consequence, bias in AI systems can compound inconspicuously.

Consider, for example, the critical role of ML (machine learning) developers in this pipeline. They are asked to: preprocess the data appropriately, choose the right models from several available ones, tune parameters, and adapt model architectures to suit the requirements of an application. Suppose an ML developer was entrusted with developing an AI model to predict which loans will default. Unaware of bias in the training data, an engineer may inadvertently train models using only the validation accuracy. Suppose the training data contained too many young people who defaulted. In this case, the model is likely to make a similar prediction about young people defaulting when applied to test data. There is thus a need to educate ML developers about the various kinds of biases that can creep into the AI pipeline.

Defining, detecting, measuring, and mitigating bias in AI systems is not an easy task and is an active area of research.4 A number of efforts are being undertaken across governments, nonprofits, and industries, including enforcing regulations to address issues related to bias. As work proceeds toward recognizing and addressing bias in a variety of societal institutions and pathways, there is a growing and persistent effort to ensure that computational systems are designed to address these concerns.

The broad goal of this article is to educate nondomain experts and practitioners such as ML developers about various types of biases that can occur across the different stages of the AI pipeline and suggest checklists for mitigating bias. There is a vast body of literature related to the design of fair algorithms.4 As this article is directed at aiding ML developers, the focus is not on the design of fair AI algorithms but rather on practical aspects that can be followed to limit and test for bias during problem formulation, data creation, data analysis, and evaluation. Specifically, the contributions can be summarized as follows:

• Taxonomy of biases in the AI pipeline. A structural organization of the various types of bias that can creep into the AI pipeline is provided, anchored in the various phases from data creation and problem formulation to data preparation and analysis.

• Guidelines for bridging the gap between research and practice. Analyses that elucidate the challenges associated with implementing research ideas in the real world are listed, as well as suggested practices to fill this gap. Guidelines that can aid ML developers in testing for various kinds of biases are provided.

The goal of this work is to enhance awareness and practical skills around bias, toward the judicious use and adoption of AI systems......' 

The Hacking of McD's Ice Cream Machine

 A little surreal, but a reminder that anything can be hacked.  I had heard of the problem from McDonalds,  Had been unaware of some of the details involved.  Just the intro below.  Quite a story.  

They Hacked McDonald’s Ice Cream Machines—and Started a Cold War  in Wired

Secret codes. Legal threats. Betrayal. How one couple built a device to fix McDonald’s notoriously broken soft-serve machines—and how the fast-food giant froze them out.OF ALL THE mysteries and injustices of the McDonald’s ice cream machine, the one that Jeremy O’Sullivan insists you understand first is its secret passcode.

Press the cone icon on the screen of the Taylor C602 digital ice cream machine, he explains, then tap the buttons that show a snowflake and a milkshake to set the digits on the screen to 5, then 2, then 3, then 1. After that precise series of no fewer than 16 button presses, a menu magically unlocks. Only with this cheat code can you access the machine’s vital signs: everything from the viscosity setting for its milk and sugar ingredients to the temperature of the glycol flowing through its heating element to the meanings of its many sphinxlike error messages.

“No one at McDonald’s or Taylor will explain why there’s a secret, undisclosed menu," O’Sullivan wrote in one of the first, cryptic text messages I received from him earlier this year.  ... '

Machine Learning Developed in Space

We have used computers in space for a long term, now a continuation of developing code there.

Raspberry Pi Heading into Space for Python Programming Challenge

ZDNet, Liam Tung, September 14, 2021

Upgraded Raspberry Pi computers will return to the International Space Station (ISS) for use in what the European Space Agency (ESA) calls the Mission Zero and Mission Space Lab challenges. Mission Zero invites coders to write a Python algorithm to take a humidity reading onboard the ISS that is shown to the astronauts with a personalized message. ESA said Mission Space Lab challenges teams of young people "to design and write a program for a scientific experiment that enhances our understanding of either life on Earth or life in space." The new Astro Pi units are Raspberry Pi 4 Model B featuring 8 GB of memory, a camera, a machine learning accelerator, sensors, gyroscope, accelerometer, magnetometer, and a light-emitting diode matrix for visual feedback. ESA said the accelerator will allow teams "to develop machine learning models enabling high-speed, real-time processing."

Wal-Mart Testing Self Driving Cars with Ford

Unexpected thing.  With a delivery angle too.. 

Walmart to Test Self-Driving Cars with Ford, Argo AI

CNBC, Michael Wayland, September 15, 2021

Walmart, the world’s largest retailer, is extending its self-driving vehicle program to include Ford Motor and Ford/Volkswagen-backed autonomous car startup Argo AI, in order to trial autonomous deliveries in Miami, the District of Columbia, and Austin, TX. Customers can place order groceries and other items online for door-to-door autonomous delivery under the program. Walmart's Tom Ward said, "This collaboration will further our mission to get products to the homes of our customers with unparalleled speed and ease, and in turn, will continue to pave the way for autonomous delivery." An Argo AI spokesperson said the partners initially will deploy a small fleet of autonomous delivery vehicles in each of the three trial cities, which will expand over time. Argo AI's Bryan Salesky said the deployments will demonstrate "the potential for autonomous vehicle delivery services at scale."

Visirule Business Risk Advisor

Have never stopped looking at simplified rule based systems to examine and  model decisions.  Some time ago we looked at some of Clive Spenser's work in this area.  Simplified and impressive.     Used similar methods in the 90s at P&G.

Storing and using complex knowledge does not have to be overly complex or too much like 'AI'.   Here it addresses risk, but can be aimed at other knowledge based applications.  Plan to test further, do take a look. Be glad to introduce you. 

Clive writes:   The BRAT initiative is based on work we have been doing for a major retail bank in the area of Testing Risk Assessment.

We simply recast it for demonstration purposes into the area of risk associated with various business activities.

The table of artifacts (Risk Areas and Risk Topics) which underpins both systems is represented using a flex frame hierarchy

The questions are standard VisiRule questions and there's a handful of statement boxes which look at the answers and determine which Risk Areas are relevant, set their initial priority levels and calculate some risk levels.

The calculated risk levels are used to calibrate the priority levels for each Risk Topic within a triggered Risk Area. Different Risk Areas are triggered by the various Business Activities.

You can play with the demo on:

https://visiruleexamples.com/visirule/tables/webflex.exe?webflex=business_risks

So the whole thing is pretty configurable and runs on our VisiRule/Flex/Prolog AWS web server using IIS/CGI  ...   

Regards,     Clive Spenser,  LPA VisiRule,  www.visirule.co.uk,    www.lpa.co.uk