1.6 Overview of the Thesis
- Chapter 2 - Related Work — We survey, identify, and describe related
work for leveraging concept languages for information access.
- Chapter 3 - Experimental Methodology — The basic building blocks
pertaining to the evaluation of information retrieval experiments, the test
collections we use in the thesis, and the setting of various parameters are
- Chapter 4 - Query Modeling Using Feedback Information — We look at
and evaluate various query modeling methods for relevance feedback in the
context of generative language models. We explicate the relation between
two popular models and introduce two novel methods that estimate a
query model using information from each feedback document individually
and combined. While most previous approaches focus either on features of
the entire set or of the individual relevant documents, our models exploit
features of both.
- Chapter 5 - Query Modeling Using Concepts — We then turn to using
concept languages to estimate a query model. In this chapter we propose
generative concept models as an extension to query modeling within
the language modeling framework, which leverages manual document
annotations using controlled vocabularies to improve retrieval. By means
of relevance feedback the original query is translated into a conceptual
representation, which is subsequently used to update the query model.
- Chapter 6 - Linking Queries to Concepts — Next, we take a closer look at
identifying relevant concepts with respect to a user’s query. In the previous
chapter we used existing document annotations and relevance feedback to
obtain concepts for queries. In this chapter we look at how we can apply
supervised machine learning models to this task and compare it to several
baseline methods including a straightforward lexical match and a purely
retrieval based approach.
- Chapter 7 - Query Modeling Using Linked Concepts — In this chapter
we bring techniques from the previous chapters together. We apply the
supervised machine learning method presented in Chapter 6 to queries
associated with two web-scale test collections. We link each query to
Wikipedia articles and apply the ideas presented in Chapters 4 and 5 to
estimate a query model.
- Chapter 8 - Conclusions and Future Work — Here we summarize our
contributions and describe potential areas for future work.
Chapter 2 and Chapter 3 serve as introductory chapters to the field of information
retrieval, language modeling for information retrieval, mapping free text to structured
knowledge sources, and experimental evaluation in the context of information retrieval.
We recommend that the reader first get familiarized with the material presented there
before reading other chapters. Many of the contributions made in the thesis converge in
Chapter 7 and to be able to appreciate the results presented there, we encourage
the reader to start with earlier material, in particular with Chapter 4 and
Chapter 6. In appendix A (See page 477), we include a nomenclature and list of