Fredrik Olsson's Licentiate of Philosophy Thesis:

Requirements and Design Considerations for
an Open and General Architecture for
Information Refinement

[updated 7:29 100515]
agenda: onsite links - offsite links


On this page, you'll find information about my Licentiate of Philosophy thesis which I defended on March 13, 2002, in Uppsala. The presentation is available in Swedish.

Abstract

This thesis presents a requirement analysis and a design proposal for a general architecture for a specified, yet open set of language engineering tasks. The chosen set of tasks is information refinement.

The need for general and reusable software for language engineering is widely acknowledged by the industry as well as by the research community. But it is hard, if not even impossible, to specify and implement software that is general enough to fulfill all possible needs that industry and researchers may have. There is a number of challenges, varying along several dimensions, that have to be taken into consideration, e.g., the language or the domain to be modelled, the characteristics of the task that the software is intended to solve, and the type of users which the software is intended to help. In aiding a developer to accomodate for these challenges, an obvious measure to take is to constrain the characteristics of the tasks to make them form a set of related language engineering tasks. In order for that set to be of use, it should be small enough to facilitate the development of general and reusable software, albeit large enough to justify the overhead that is involved in developing such software.

The present work introduces information refinement as a set of related tasks intended to serve as a target for developing a general and open architecture, Kaba. The notion of information refinement involves techniques intended to grant users access to the right textual information at the right time, e.g., information extraction, information retrieval and automatic text summarisation, while taking into consideration factors such as the users' information need, their context (e.g., knowledge of the domain at hand), and their situation (e.g., work process).

The requirement analysis and design proposal presented here are formed by three parts: the notion of information refinement; a survey of a number of projects and software that have had great impact on how language engineering software is constructed today; and on the experience gained from a case study on constructing a language processing tool-set for Swedish, SVENSK.

Download the thesis

Fredrik Olsson. 2002. ``Requirements and Design Considerations for an Open and General Architecture for Information Refinement''. Licentiate of Philosophy Thesis, Uppsala University, Uppsala, Sweden, March. Available as as RUUL No. 35 (Reports from Uppsala University, Department of Linguistics). ISBN: 91-973737-1-0, ISSN: 0280-1337. NOTE: this version is formatted for printing on a5 paper
[ ps | pdf ]

Licentiate proposal text and seminar

The licentiate thesis proposal is continously revised as new issues pop up in my discussions with my supervisors. This list is in reversed chronological order, i.e., the latest version is always at the top.

The revised proposal for the licentiate thesis. October 16rd, 2000. [ps | pdf]
In Swedish, approximately 120 Kb. The proposal is revised according to the discussions on the supervision meeting October 13th.

The revised proposal for the licentiate thesis. August 23rd, 2000. [ps | pdf]
In Swedish, approximately 120 Kb. The proposal is revised according to the discussions on the licentiate thesis proposal seminar.

Slides for the licentiate thesis proposal seminar [ppt | html]
In English. The seminar was held on August 22nd, 2000. PPT version is approximately 77 Kb.

The original proposal for the licentiate thesis. May 3rd, 2000. [ps | pdf]
In Swedish, approximately 120 Kb.
maintained by fredrik olsson fredriko@sics.se