We conduct research within the field of Artificial Intelligence (AI).
AI is concerned with programs that provide
solutions (or approximations to solutions) to problems which are
'difficult' to solve by traditional methods. The difficulty stems
from the presence in the problem of disorder, uncertainty, lack of
precision or inherent intractability.
In particular, the Group is interested in knowledge-based systems for
intelligent retrieval and decision-support.
But we have also made contributions to areas such as natural language
processing, machine learning and ant algorithms.
Research questions in Case-Based Reasoning and Recommender Systems
Case-Based Reasoning (CBR) is an approach to problem solving which
can be successful in domains where "similar problems have similar
solutions". In such domains, new problems can be solved by
transferring and adapting solutions that were used to solve
similar problems in the past.
Recommender systems help us to decide which goods, services or
information to consume based on what they infer about our
preferences.
Our research addresses questions such as:
In a CBR system how do we retrieve the best cases, and
in a recommender system how do we select what to recommend?
How can these systems learn from experience, and how well do they learn?
How can these system maintain their knowledge, especially if it
is changing over time?
How can these systems explain their actions and conclusions and
advise their users?
How can these systems reason with unstructured information (e.g. text), as well
as structured information?
How can we integrate multiple reasoning engines (e.g.
case-based with rule-based,
case-bases with databases,
case-based with constraint processing,
case-based with genetic algorithms & genetic
programming,
ant algorithms with constraint processing,
collaborative filtering with content-based filtering,
etc.)?
Example application areas
We have applied our research to problems in several domains.
In spam filtering we face challenges such as handling
unstructured information,
and using learning to cope with concept drift (i.e. the fact that
spam is a moving target).
To recommend configurable products such as travel itineraries and high-end consumer
electronics requires the ability to reason not only with the user's
preferences but also with constraints and policies that determine how
well different sub-components 'go together'.
In a waste exchange system we seek to pair those who produce
waste with those who can reuse it, with the challenge that neither party
can be persuaded to describe the waste they have or want with sufficient
precision.
In predicting the product yield of the trees in a forest
we have to estimate tree stem diameter along the length of each tree from laser scanner data
that is noisy and gappy.