After having described relevance and information retrieval in general and legal context, as well as reviewing previous work in the field as well as designing and evaluating a better relevance ranking method, are we done? No, we’ve only just started! Here are some pointers on how this approach might be improved.
Finally, this is the heart and soul of this thesis (even if it’s only a few pages). A system designed for better legal relevance ranking is described and evaluated. Although primitive, being based only on simple known link analysis algorithms, it seems to perform really good compared to traditional ranking methods.
There have been many attempts to improve relevance judgments in legal information retrieval systems, with many approaches. This chapter describes some of these, to better understand the context in which the prototype system described in the next chapter appears.
Relevance can be interpreted in many ways, from subjective to objective. Which interpretations are built into traditional information retrieval systems, and what properties does these manifestations of relevance have? The use of IR for legal information has a long history. How does legal information retrieval correspond to the legal method, and can we improve on this correspondance, by e.g. creating a relevance ranking function more in line with what is considered legally relevant?
In order to define a better relevance ranking method, we need to delve deep into what relevance really is, and what aspects of it we can measure in an information retrieval system. We also examine what relevance means in a legal context, and how it is connected to other concepts such as authority and what clues to relevance we can find in legal information.
The first chapter sets the scene by describing the basis for information retrieval systems, legal information and how it is used, as well as the motivation for improving the former so that we can use the latter better. It also contains a description of the method used in the thesis, as well as the general structure of it.
My graduate thesis, somewhat loftily titled ”Towards a theory of jurisprudential relevance ranking – Using link analysis on EU case law” has been submitted to and approved by my supervisor. It has taken far too long time since I first started working on it, but I’m very satisfied that it is finally finished. Except that it’s not really finished, since I hope to re-work and extend it with the aim of publishing it in some other form. Which is why I’m soliciting feedback on it.
Over the coming week, I’ll be publishing a chapter at a time. Each chapter will be available in PDF form and also inline in the form of images. This since that was the best conversion to a web-friendly format I could manage… (also note that the pagination differs slighly between the PDF and the web version).
If you are at all interested in legal informatics, information retrieval, jurisprudence or just what we really mean when we say that something is relevant, I hope you will find the time to read the chapters and maybe also give me your feedback below.
We’ll be kicking of with the front matter of the thesis. It does not contain anything substantial in itself, but it has a very neat Gephi-drawn cover and some interesting quotes. The table of contents should give you an idea of what it is about.