Internship | Sherlock Holmes in the age of AI

Problem
Sherlock Holmes is famous for being able to craft theories, hypotheses and conclusions from the evidence he encounters at crime scenes. During investigations in which he does not have all the information required to determine the suspect or the murder weapon with absolute certainty, he needs to reason about plausible causes for what he observes. This sort of abductive reasoning, the ability to reason from observations to explanations and deal with missing information is a fundamental ability of human thought processes. Examples in other domains where diagnosis is required include inferencing from a set of symptoms to a disease (medical diagnosis), or from a malfunction in a system to a faulty part in a system (model-based diagnosis).

Increasingly, we will deploy AI systems that collaborate with a human to provide assistance in the diagnosis process. Like Sherlock, these AI systems will encounter situations where they have limited data about the state of the environment: the problems of noise or missing observations always exist in real world applications. AI systems must then still be able reason to "the best explanation" or generate a hypothesis for a given situation, based on evidence available to aid humans.

Much of the knowledge we require for our reasoning processes, is formalized in ontologies and Description Logic (DL) as a means of representing domain knowledge. This logic allows reasoning from cause to conclusion (deductive) and does not naturally support abductive reasoning. There are several approaches that have investigated abductive reasoning in a knowledge base based on DL, but they do not yet present a sufficiently practical approach for real-world settings and require further research. In addition, previous work on this topic has mainly looked at procedures for generating hypotheses without human involvement. In a collaborative setting, we can foresee the AI system interacting with a human to refine its reasoning process and its hypothesis, by e.g. explaining how the hypothesis is constructed or requesting specific information to refine its theory.

Goal
The goal of the research is to continue our group’s research in abductive reasoning in a knowledge base of DL, by demonstrating human-assisted reasoning based on a use case ontology. This research will be part of the Hybrid AI research program within TNO.

Solution
At TNO we work with ontologies in various domains and currently employ deductive reasoning on knowledge bases in our Plasido knowledge base platform (based on Apache Jena). This platform needs to be extended with a Sherlock component that supports abductive reasoning on ontologies in a human-assisted setting. A basic implementation of an abductive reasoner is currently available.

Method
This thesis will consider existing literature, tools and approaches to human-assisted abductive reasoning in ontologies. The student will then design a framework for human-assisted hypothesis generation. Subsequently, the theory will be demonstrated and implemented in a reasoning component, as an extension of our Plasido platform. The student will validate its implementation via an abductive reasoning process in a use case.

Your background and interests should include knowledge representation, formal logic and programming. We expect you to be imaginitive and creative in thinking about new solutions and innovations that are out-of-the-box and not yet invented for the problem at hand. In addition, you are expected to be a teamworker who is also interested in the work of others in the department and who has an open eye for related work that can be applied to the assignment. Finally, a result-oriented attitude is necessary for this assignment and it is expected that a working prototype of the proposed solution will be realized.

TNO is an independent research organisation whose expertise and research make an important contribution to the competitiveness of companies and organisations, to the economy and to the quality of society as a whole. Innovation with purpose is what TNO stands for. With 3000 people we develop knowledge not for its own sake but for practical application. To create new products that make life more pleasant and valuable and help companies innovate. To find creative answers to the questions posed by society. We work for a variety of customers: governments, the SME sector, large companies, service providers and non-governmental organisations. Working together on new knowledge, better products and clear recommendations for policy and processes. In everything we do, impact is the key. Our product and process innovations and recommendations are only worth something if our customers can use them to boost their competitiveness.

You want to work on the precursor of your career; a work placement gives you an opportunity to take a good look at your prospective future employer. TNO goes a step further. It’s not just looking that interests us; you and your knowledge are essential to our innovation. That’s why we attach a great deal of value to your personal and professional development. You will, of course, be properly supervised during your work placement and be given the scope for you to get the best out of yourself. Naturally, we provide suitable work placement compensation.