Semantic Desktops

Description

Our work focus on supporting situation assessment for a joint group of military mission planners. In their work, mission planners work with a vast array of different tools and data sources. They communicate via radio, electronic messaging, e-mail and documents. They manage geographical information through specialized GIS systems and access sensor equipment through yet other systems. In all, this environment puts high demands on commanders and their information processing abilities.

Semantic desktops present users with a coherent view of their information based on concepts from their work context rather than the metaphors dictated by the desktop environment. For example, a user may select to view all people he or she has any connection to, which may include co-authors of documents, contacts in instant messaging programs and e-mail recipients, all in a unified view. The semantic desktop vision also includes a vision of common access to all semantic information used by a group of people, so that all mission commanders in a group should have shared access to information about what others are doing.

Currently, we are working with a Semantic Desktop environment called OpenIRIS which enables users to reason about and relate concepts in their work environment while preserving as much as possible of their natural tools and behaviour.

Demos

Managing communications effectively is imperative for efficient staff work. With the help of a semantic desktop that collects information about events such as incoming or outgoing e-mail, we have created a tool to visually present and analyze communications within a staff environment. The demo below shows how the communication between a student, a supervisor and an examiner follow the prescribed workflow of an undergraduate thesis project. The transitions between different states in the work flow are recognized by e-mails sent between the different actors.

• Workflow analysis

Student projects

We accept undergraduate students who are willing to work on projects as a part of their undergraduate education or as degree projects. The topics listed below are those we are primarily interested in, with a rating in parenthesis indicating the estimated difficulty:

• (2) Implementation of Open Document support for the natural language processing tool GATE, to facilitate the analysis of the contents of OpenOffice documents.
• (3) Implementation of a two-way mapping between the marker position of an OpenOffice windows and the corresponding position of the document file. This mapping should be used to create a tool for defining semantic domain entities in documents by marking regions of an open document.
• (1) Integration of Google Maps in the OpenIRIS semantic desktop environment so that map-related ontology information can be displayed with Google Maps markers.
• (2) Visualization of communication between members of staff by creating an OpenIRIS plugin that acts on events corresponding to receiving and sending e-mail messages.

Thesis projects

We offer theses for Masters students who have a strong technical interest, and a solid understanding of programming in larger projects. If you have successfully completed one of our advanced programming courses such as TDDB69 or TDDD27, you probably have what it takes. For a more detailed overview of our research field and research questions, see the thesis description overview.

A workflow critiquing application for OpenIRIS

The IRIS Semantic Desktop environment keeps track of a number of applications and the data produced in them. This information can be used to produce visual presentations of data, to merge different data sources and present relationships between data in a general sense. Until now, however, the data gathered has not been used to give directed feedback and support on the work of users. In this project, you will create an addition to the OpenIRIS environment in which users can describe their expected work flow in terms of what people are responsible for what information, who should produce certain documentation and what people need information. Based on this work flow description, the tool should be able to perform an analysis on the work performed by a single user or a groups of users and determine of the prescribed work process is followed.

Context editor

An important part of creating a domain-specific Semantic Desktop application is the possibility to define the context of the domain. The context is in turn defined by the vernacular used, the structure of documents and messages, and the established work process. By providing users with a means of specifying their work context in a unified manner, a Semantic Desktop environment can become a much more potent support. This thesis focuses on how to model and describe semantic entities described by either structure or content as an end-user. Describing these entities also involves establishing a connection between item occurrences in messages or documents and classes in the IRIS knowledge base.

Publications

[1]

Ola Leifler and Henrik Eriksson. Domain-specific knowledge management in a semantic desktop. In Proceedings of I-KNOW '09, The International Conference on Knowledge Management, 9 2009. [ bib ] Semantic Desktops hold promise to provide intelligent information-management en- vironments that can respond to users' needs. A critical requirement for creating such environ- ments is that the underlying ontology reflects the context of work properly. For specialized work domains where people deal with rich information sources in a context-specific manner, there may be a significant amount of domain-specific information available in text documents, emails and other domain-dependent data sources. We propose that this can be exploited to great effect in a Semantic Desktop to provide much more effective knowledge management. In this paper, we present extensions to an existing semantic desktop through content- and structure-based informa- tion extraction, domain-specific ontological extensions as well as visualization of semantic enti- ties. Our extensions are justified by needs in strategic decision making, where domain-specific, well-structured knowledge is available in documents and communications but scattered across the desktop. The consistent and efficient use of these resources by a group of co-workers is crit- ical to success. With a domain-aware semantic desktop, we argue that decision makers will have a much better chance of successful sense making in strategic decision making. Keywords: semantic desktop, knowledge management, domain-specific ontology

[2]

Ola Leifler and Henrik Eriksson. A model for document processing in semantic desktop systems. In Proceedings of I-KNOW '08, The International Conference on Knowledge Management, 9 2008. [ bib ] Keywords: knowledge management, information extraction, semantic desktop, command and control