Data Mining and Machine Learning are in the midst of a "structured revolution". After many decades of focusing on independent and identically-distributed (iid) examples, many researchers are now studying problems in which examples consist of collections of inter-related entities or are linked together. A major driving force is the explosive growth in the amount of heterogeneous data that is being collected in the business and scientific world. Example domains include bioinformatics, chemoinformatics, transportation systems, communication networks, social network analysis, link analysis, robotics, among others. The structures encountered can be as simple as sequences and trees (such as those arising in protein secondary structure prediction and natural language parsing) or as complex as citation graphs, the World Wide Web, and logical knowledge bases. In all these cases, structured representations can give a more informative view of the problem at hand, which is often crucial for the development of successful mining and learning algorithms.

The field of {\it probabilistic, logical and relational learning} (aka. {\it statistical relational learning}, {\it probabilistic inductive logic programming}) tackles the structured input-output problem sketched above by combining expressive knowledge representation formalisms such as relational and first-order logic with principled probabilistic and statistical approaches to inference and learning, and hence lies at the heart of artificial intelligence. It is a relatively young and all the more active field of research offering a lot of research opportunities. This was already witnessed by a previous seminar on "Probabilistic, Logical and Relational Learning - Towards a Synthesis" that took place from January 30 to February 04, 2005, which succeeded in bringing together a significant number of researchers from all over the world that are working on all aspects of probabilistic, logical and relational learning. The result was a better understanding of the common grounds of this newly emerging field and the identification of a number of key research challenges.

The goal of the 2007 seminar was to provide answers to some of this key research challenges in the area, including:

• What is the relationship among the many different probabilistic, logical and relational representations that are being used?
• What are suitable settings for learning such representations? And, what are the challenges raised by the different learning settings? Also, can one arrive at a learning theory focused on probabilistic, logical and relational representations?
• What are the application areas for which probabilistic, logical and relational learning is well-suited? What does it take to develop show-case applications in these areas? Can we identify common and concrete application challenges on which progress can be measured and techniques? Providing answers to these questions should --- ultimately --- provide the field with a commonly agreed upon framework as well as provide an application focus, which together could form the basis for further developments in the area.

Not all of the questions could have been answered yet but significantly progress has been made as shown by the great collection of abstract below. They have been collected from 45 seminar attendees from 11 different countries. The presentations at the seminar, varying in length, covered a large variety of topics, including novel results on lifted inference within first-order probabilistic languages, learning infinite relational models, statistical predicate invention, and applications within citation analysis, robotics, and life sciences. Talks were spread over the week to allow for plenty of time for discussions. Breakout sessions on special interest topics were organized on the fly using the Seminar's Wiki page. The breakout sessions gave the participants a chance to exchange problems and discuss ideas and challenges lying ahead indepth. We are positive that many of the breakout sessions will lead to new results, collaborations, and publications. Within the talks and the breakout sessions, we saw very lively debates showing the growing demand and opportunities for statistical relational learning within theory and practice of machine learning. We were also very pleased to see the significant progress made between the present seminar and the previous one. This was very clear in the demonstration session, where a number of academic prototypes of probabilistic, logical and relational learning systems were presented.

As usual, Schloss Dagstuhl proved to be an excellent place to hold a great meeting, so we would not only like to thank the participants of the seminar for making this a very successful event, but also the Dagstuhl staff for providing a friendly and stimulating working environment. Finally, we would like to thank Sriraam Natarajan for his valuable help in collecting the abstracts and full-text contributions.