Understandability is a key challenge in explanation provision and estimating well what users may already know can help offering better explanations.

Abstract. The research project MEDICO aims at developing an intelligent, robust and scalable semantic search engine for medical documents. The search engine of the MEDICO demonstrator RadSem is based on formal ontologies and is designated for different kinds of users, such as medical doctors, medical IT professionals, patients, and policy makers. Since semantic search results are not always self-explanatory, explanations are necessary to support requirements of different user groups. For this reason, an explanation facility is integrated into RadSem employing the same ontologies for explanation generation. In this work, we present a user experiment that evaluates the intelligibility of labels provided by the used ontologies with respect to different user groups. We discuss the results for refining our current approach for explanation generation in order to provide understandable justifications of semantic search results. Here, we focus on medical experts and laymen, respectively, using semantic networks as form of depiction.
[Björn Forcher, Kinga Schumacher, Michael Sintek, and Thomas Roth-Berghofer. Evaluating the intelligibility of medical ontological terms. In Joachim Baumeister and Grzegorz J. Nalepa, editors, Proceedings of the 5th Workshop on Knowledge Engineering and Software Engineering (KESE-2009), http://CEUR-WS.org/Vol-486/.]

Participants of ExaCt 2009 on day #1

The ability to explain reasoning processes and results can substantially affect the usability and acceptance of a software system. There is no doubt about it. Unfortunately, the topic of explanation has not received proper attention since the demise of expert systems research during the “AI winter”. Only recently explanation is seen as a research topic in its own right again. At my last workshop on Explanation-aware computing ExaCt 2009, organised and held together with Nava Tintarev and David B. Leake at IJCAI-09 in Pasadena, CA, several more renowned researchers joined my effort of pushing and promoting the topic of explanation; among them Deborah McGuinness (RPI, USA), Anind Dey (CMU, USA), Ashok Goel (Georgia Tech, USA), Doug Walton (U. of Windsor, CA), and Miltos Petridis (U. of Greenwich, UK). (More researchers can be found on my dedicated website on-explanation.net.)

So, what do I want to achieve?

First, I want to make software designers and developers aware of explanations. In the long run, I also want software systems to become explanation-aware. I want to promote the notion of explanation as a research topic in its own right in order to develop respective engineering methods.

How to get there

A lot of inspiration came from Edward Tufte’s book Visual Explanations, which I stumbled upon by accident on the desk of a former colleague, years before I took an active interest in explanation. Edward Tufte invites the reader to

enter the cognitive paradise of explanation, a sparkling and exuberant world, intensely relevant to the design of information.

As Artificial Intelligence is about simulating human intelligence, we AI researchers should take Tufte’s words to heart. In AI, we strive—not only but also—for the goal that AI systems become able to discover explanations themselves and that they represent them appropriately in order to communicate with their users. Until that goal is reached, we should at least provide such systems with pre-formulated explanations and representation templates to support human users in their interaction with the system.

Within the field of knowledge-based systems, explanations are considered as an important link between humans and machines. Their main purpose is to increase the confidence of the user in the system’s result (persuasion) or the system as a whole (satisfaction), by providing evidence of how the solution was derived (transparency). Explanations are part of human understanding processes and part of most dialogues, and, therefore, need to be incorporated into system interactions. But looking at all the efforts already invested in explanation research, I think, we have just rattled at the gates of the above mentioned cognitive paradise.

A helpful tool for designing and developing software systems from an explanation-aware viewpoint is the general explanation scenario depicted here with three participants: user, originator, and explainer. The user communicates by way of a user interface (UI) with the whole software system and is the recipient of explanations. The originator is the tool the user works with to perform tasks and solve problems. The explainer can be seen as another tool that helps understanding how the originator works and what knowledge the originator uses. Please note that this scenario is simplifying in so far as it does not consider the case where the software system asks the user for explanations or justifications.

Explainer and originator need to have knowledge about each other. The originator needs to provide knowledge about its reasoning process as well as intermediate results and decisions in order to allow the explainer to generate good explanations. The relationship between explainer and originator is somewhat asymmetrical regarding the used knowledge. Whereas the explainer needs to have access to the originator’s knowledge-base (in addition to its explanation supporting knowledge-base) this does not hold for the originator. For its problem-solving task the originator does not need to have access to the explainer’s knowledge-base, but the originator needs to be aware of it in order to fill it appropriately.

Community building

Making people—researchers, software designers and developers, and, not to forget, funding organisations—aware of explanation-awareness (pun intended) is not a one-man show but a community effort. At last year’s ExaCt 2008 the idea of a manifesto came up as a way of expressing goals of the developing explanation community. A first version is available here.

If you would like to participate in further discussions or just like to receive further information on this topic and about future workshops you might consider joining the Yahoo!-group explanation-research. You think you or someone else belongs on the list of explanation researchers? Drop me a line and a URL to their  homepage and indicate relevant research.

Update: Due to the lack of enough submissions the workshop needed to be cancelled.

The human is always at the center of my research. So it is no surprise that I’ll try to investigate another facet of socio-technical computational systems in the upcoming workshop on Human Centered Case-Based Reasoning HCCBR. The workshop is part of the 8th International Conference on Case-Based Reasoning ICCBR 2009 and will take place 20–23 July 2009, in Seattle, Wahington. Here is an excerpt of the call for papers (submission deadline: 23 March 2009):

Human-centered computing focuses on methodologies and technologies to improve the interaction and performance of socio-technical systems. Intelligent systems are not longer considered to be black boxes that provide a full solution to a problem on their own, instead, problem solving is seen as an interactive process. Case-Based Reasoning would appear to be a natural fit for such integrated human/computer systems.

A number of important questions are raised by Human-centered approaches. It is important to gain a better understanding of how each part of the combined system can help to extend the capabilities of the other. It is also of interest to examine how human knowledge modelling and construction can best be supported through technology.

Many of these issues are already being addressed in other disciplines, however, the question remains as to how findings from the social sciences and psychology may be integrated into the design of CBR systems. This integration ranges from psychologically plausible knowledge models to the delivery of an attractive end user experience.

Another important research topic in human centred computing centres around the issue of communication and, in particular, explanations. Problem descriptions, as well as other input, can be incomplete and changing. As a consequence, there has to be communication between human and software agents. Communication requires mutual understanding that can be essentially supported by explanations.

Further, context sensitive processing plays a key role in many modern IT applications. Context-awareness and context-based reasoning are essential not only for mobile and ubiquitous computing, but also for a wide range of other areas such as collaborative software, web engineering, personal digital assistants, information sharing, health care workflow and patient control, adaptive games, and e-Learning solutions.

From an intelligent systems perspective, a further challenge is to integrate context with other types of knowledge as an additional major source for reasoning, decision-making, and adaptation and to form a coherent and versatile architecture. There is a common understanding that achieving desired behaviour from intelligent systems will depend on the ability to represent and manipulate information about a rich range of contextual factors.

This workshop aims to bring together researchers and practitioners exploring human aspects of the design, implementation, and use of intelligent systems, from a broad range of areas, to share their problems and methodologies across different research and application areas. The workshop will examine methods, mechanisms, and techniques to keep the human in the centre of attention during the whole lifecycle of an intelligent system, from initial problem description through to knowledge acquisition and modelling and on to interactive use and maintenance.

Please consider submitting a paper!

Already a month ago I enjoyed very much going to the 28th International Conference on Artificial Intelligence AI 2008 and the UK Case-Based Reasoning workshop UKCBR 2008. The workshop was the thirteenth of this successful series. It was the first time I went to either event and I intend to go again. That they took place in Cambridge at Peterhouse college was an added bonus. You can find some photos here.

I had my hands full with four presentations over the course of the three days.

At UKCBR I presented the “Explanation Capabilities of the Open Source Case-Based Reasoning Tool myCBR” (co-authored with my student Daniel Bahls). I am also happy that Miltos Petridis, organiser of UKCBR invited me to demonstrate the rapid prototyping capabilities of myCBR in a demonstration session.

A special highlight was the invited talk “Using Statistical Translation Models for Textual CBR” by Luc Lamontagne from Laval University, Québec, Canada.

At AI-2008 I gave two talks, both related to my explanation research. The presentation “Code Tagging and Similarity-based Retrieval with myCBR”, also co-authored with Daniel, presented the coTag system. coTag is a code tagging plug-in that allows annotating code snippets in the integrated development environment eclipse. It offers an easy-to-use interface for tagging and searching not only for exactly the same but for similar tags. For better understanding how coTag comes up with its solution we used the above mentioned explanation capabilities of myCBR.

My second talk was based on a paper I wrote with another student, Florian Mittag: “ReduxExp: An Open-source Justification-based Explanation Support Server”. Here we worked on a mechanism for recording decisions of information systems. In the talk I described the architecture of ReduxExp and first experiences with the system.

Highlights of AI 2008 surely were the machine intelligence competition with live presentations (strictly limited to 15 minutes each) and the Gala Dinner held by candlelight (and without enough heating!) in the dining hall of Peterhouse college.

“The Loebner Prize for artificial intelligence ( AI ) is the first formal instantiation of a Turing Test. The test is named after Alan Turing the brilliant British mathematician. Among his many accomplishments was basic research in computing science. In 1950, [...] Alan Turing asked the question “Can a Machine Think?” He answered in the affirmative, but a central question was: “If a computer could think, how could we tell?” Turing’s suggestion was, that if the responses from the computer were indistinguishable from that of a human, the computer could be said to be thinking. [...]

In 1990 Hugh Loebner agreed with The Cambridge Center for Behavioral Studies to underwrite a contest designed to implement the Turing Test.”

[From Loebner Prize for Artificial Intelligence]

This summer I was involved as a preliminary phase judge (Loebner Prize 2008) and enjoyed testing some of the systems. I am not entirely sure about the value of the test itself, but I am sure that these kinds of contests help improving interactiveness of software systems.

Last week I attended the ninth European Conference on Case-Based Reasoning ECCBR 2008 in Trier, Germany. Just around the corner of Kaiserslautern, one could say.

The conference was again a great experience. The community is special. I always feel welcome and at home. From talking to other participants to this conference series I know that even people who attended the conference for the first time feel the same.

A first highlight of this year’s conference was the invited talk by Pádraig Cunningham and Barry Smyth. They analysed research themes in CBR conference literature. One motivation was to check whether our often pessimistic view of our community was founded. It was quite nice to learn that they found that the CBR community is in fact quite healthy. What a relief Their analysis discovered a few evolving CBR themes such as recommender systems and diversity, case-base maintenance, adaptation, creativity and knowledge-intensive CBR, and conversational CBR. Their evaluation shows clear evidence of sustained innovation and maturing research. An important discovery surely is that the impact factor is comparable with big AI conferences such as ECAI and ECML.

ECCBR 2008 also presented an innovation: The Computer Cookery Contest CCC. The task:

Write your own software application for the live competition. Show that your program is more creative than the average kitchen user. Let your computer’s recipe creations be evaluated by a professional cook and an international jury of scientists! [From the call for participation]

The competition was quite entertaining. Some of the systems provided, well, interesting recipes when they needed to change ingredients in order to accommodate for ingredients at hand / in the fridge. You can find the results here.

There were also two personal highlights: I was invited to participate in a panel discussion on open source software and I had a talk together with Armin Stahl where we presented myCBR for rapid prototyping and also its explanation capabilities.

The discussion panel was chaired by Ralph Traphöner, a former colleague. Mehmet Göker, with whom I have organised ECCBR 2006, took the view of a global player in software use whereas Eyke Hüllermeier, professor at University of Marburg, took on the role of academia. They discussed with three open source software providers, i.e., with Christian Brockmann (eclipse project smila), Pedro Gonzalez Callero (jColibri), and me (myCBR).

The talk about myCBR was received very well. Over the course of the conference I was asked to show the tool to several attendees who either planned to evaluate myCBR in commercial settings or want to use it for teaching purposes. In my part of the talk I presented some of the explanation capabilities built into myCBR. Go and check out myCBR

The recent death of the science fiction author Arthur C. Clarke reminded me of one of my favourite quotes:

“Any sufficiently advanced technology is indistinguishable from magic.” (Third of Clarke’s three “laws” of prediction.)

Modern information technology is for many people—and sometimes even for me—indistinguishable from magic. You cast a spell, i.e., interact in a specific way with some user interface, and if you waved your magical wand the right way—and only then—you achieve your goal. Such devices as the Nintendo Wii video game console or the iPhone with their motion detection capabilities allow for completely new interactions with the user and more natural interactions among users.

But there is a big problem with magic: You are required to believe in it and to not ask questions about it. Magic’s dark side is all about hiding, making believe, obscuring, and blinding. I do not say that software developers intentionally engineer obfuscating applications, but from a user’s point of view it often just looks the same.

Of course, there are times where I suspend my disbelief, where I need to suspend my disbelief. Every time I watch a movie or read a novel I am required to do so in order to be entertained. But life is (unfortunately?) not only entertainment. Computer systems need to have beautiful and elegant, easy-to-use interfaces that evoke a sense of wonder like magic does. This just helps keeping your spirits up and makes working more fun. There is no need for dull and boring applications. But interfaces should never be shallow. Systems should help people learn about what is going on in a software system if people want to know. Systems should provide transparency when asked for. Systems should explain their used vocabulary when asked about.

So, my dear developers, keep on providing magical interfaces, but let the interested users have a peek behind the curtains whenever they want.

When I plan a workshop I always look for interfering deadlines. This time I overlooked an obvious deadline, even though this deadline—Call 3 for IST project proposals in the seventh research framework programme (FP7) of the European Union—was also very important to me. I will blog about my proposals later.

Well, as my third workshop on Explanation-aware Computing ExaCt 2008 (see also my blog entry here) is affected by the FP7 deadline I extended the submission deadline to 16 April 2008.

One of my favourite quotes from the British TV show Little Britain is this one:

Even though this is a different context programs often just do that, they say ‘No’.

Comics always were a part of my life. I still love comics, even though my time for them is scarce. I grew up reading tons of comics at my cousin’s place, which was easily reachable via our backyard. I spent a lot of time there with Batman, Superman, The Fantastic Four, and many more super-heroes. Of course, at that time comics were seen as a very bad influence on impressionable young minds; at least as bad as computer games today, I suppose. As my cousin also drew comics herself I learned already back then a lot about the structure of comics and what makes super-heroes tick. But it took me over 25 years to actually learn some more about comics. Scott McLoud opened my eyes in his book Understanding Comics: The Invisible Art about a lot of things. This book shows and not only tells, as it uses the same medium it talks about: it itself is a comic book.

Reading Understanding Comics I was astonished how many similarities I found between comics and ontologies, i.e., formal representations of the world. Comics and ontologies represent, each in their own way, views of the world. Comics exaggerate physical and mental abilities, distort laws of nature, and portray extreme characters. Granted. But all is grounded in our own experience and would not work without our experience. For example, often action in comic book scenes takes place only in between the panels, i.e., in the mind of the reader. The artist draws the scene before and after the action takes place, and the reader fills in the blanks. And a kid surely fills in the blanks differently than an adult. I was not aware of my own interaction with the comic.

Comics abstract from the world. They are most of the time intentionally not as photo-realistic as they could be. By stripping away details the artist can let the reader focus on what is important. Scott McLoud presents several techniques used in comics, of which I knew only a few. I was not even aware of most of the others even though I would have argued that I have read comics carefully. Read, yes, but, obviously, not studied carefully.

In another chapter, Scott McLoud addresses upcoming comic artists. He presents a six step process of how to become a comic artist. Again, I was reminded of how to become a knowledge engineer. One needs to master a craft as well as have some talent. Any kind of artist has to decide in which form an idea will be expressed, following some school of art (idiom). Then the artist composes the work, leaving some things in and others out. For crafting the art, the necessary skills and knowledge need to be available. Finally, without some production skills the surface of the art work will not be appealing enough to sell the art work as the surface is the first thing a customer will base the buying decision on. Well, how come that ontologies do not sell, one could wonder.

There are a lot more interesting things to say about this book. Time can flow quite differently in comic books. The triangle of reality, picture plane, and language is discussed in detail. And much more …

After reading Understanding Comics I am able to enjoy comics on an additional level, now having more knowledge about the craftsmanship of comic artists.

I now wonder whether and how some of the techniques could be put to use in developing explanation-aware systems. Any suggestions?