School of Computer Science. The University of Birmingham

Robert Hendley

Student Projects 2011/12

Most of these projects are available to either undergraduate or MSc students (if not, then there is a note).

Most of these projects are best suited to students who would expect to get a IIi or I degree. They could be done by other students, but in many cases there are technical issues which might cause problems for weaker students.

Visualisations of Complex Systems

A wide variety of application domains exist which would benefit from an effective visualisation. These projects will build and evaluate 3D visualisations of the following domains:

· Email communications

· Search Engine results

· File systems/Desktop integration

· Non-governmental organisations (NGO) information and structure

· Business structure (personnel, departments, etc) and processes

The Emotional Network

Here the aim is to extract data from a system such as Facebook and then use the information within that to build a representation of the emotional state. This could be used to describe the emotional state of different sub-networks or to describe the emotional state over time – e.g. by time of day or after major events

The Intelligent Mobile

Using data that is now available on many ‘phones (e.g. position, proximity, motion, activity, diary, addressbook etc.) what can we infer and how can this adapt the behaviour of the ‘phone? For instance, can we infer that someone is in a meeting and then filter calls so that only urgent ones cause the ‘phone to ring? Can we infer that someone is on a busy street and then increase the volume if an important call is received?

Emergence

How do cells grow and specialise? How do ants coordinate their behaviour? There are many examples in nature of complex emergent behaviour deriving from interactions between very simple components. How can we use these techniques to better solve real problems e.g. Building better visualisations or social networking systems?

Sonification

This project is aimed at exploring the use of audio as a tool for understanding complex data (maybe text documents or even quantitative data). It might be a combination with some visualisation tool.

Annotation Projects

These are open ended projects which are focused around the idea of building automated or collaborative annotions.

· GWAP (Games With A Purpose). The aim here is to turn the annotation task into a game that will challenge the annotator to more deeply explore the space of annotations so that they properly discriminate between competing descriptions (whereas they presently will come up with descriptions that are routine). It will involve building a web-based multi-user game framework and implementing a series of alternative approaches (ie. Different ‘game’ types or competitive/competitive models)

· Challenging users. The aim here is similar to the previous project but with an intelligent agent that will reason over the annotation space to devise questions or tasks for the user that will force them to consider aspects that they had previously overlooked.

· Extracting ‘mood’ from articles. Although most annotation systems focus on concrete information it is apparent that searches by journalists are often more likely to focus upon subjective information (I need a clip that includes wind farms with a dark or negative aspect). This project will focus upon the extraction of these more subjective features from video or text.

· Annotation by analogy. Rather than annotate from scratch, how can we re-use information that is already available in similar or dissimilar articles? E.g. This article is very like this one in this area but unlike this one or this class. How can we enable the user to explain these relationships?

· Visualisation of articles. Here we are interested in allowing a user to use a visualisation of the annotations. This may be during searching or within the annotation process. It is likely that that this will use a force based clustering model or an ‘organix’ like model based on our earlier work

· Whereas we can scan a text document very quickly to see assess its content or relevance, we cannot do this for video. We can judge in seconds whether a large document or even a book would be useful—How can we do this for video material?

GoogleMaps & fault reporting/local history/Local democracy/...

Imagine you are doing a local history project or want a system to allow you to report and monitor problems in your neighbourhood (e.g. graffiti or potholes or just council stupidity). Perhaps you are working on green technologies and want to record who is doing what and where.

How can you do this? You need a fusion of googlemaps/wikis with something like a CRM system to manage and track things. In this project the aim is to take one of these areas (e.g. A history of Birmingham canals) and use appropriate APIs to provide a customised and integrated view onto these different tools.

Computer Systems for disabled users

The problems vary from difficulty in the use of conventional computer systems (keyboards, mice, typical displays) through deeper cognitive problems (poor short term memory, difficulty in generating words etc)

The aim of these projects is to develop a series of prototype systems that address the needs of these specific users. This might involve the development, use or adaptation of special input devices through to special tools to help users perform particular tasks (e.g. letter writing).

Mechanical Turk meets Price Comparison

The aim of this project is to use the idea behind the Mechanical Turk for a Price Comparison application – that real people search for the bargain. This will involve building a fairly sophisticated set of web applications to manage the requests and to dispatch tasks and collate results.

Collaborative 3D environments

The widespread use of networks and the rising impact of 3D technology make cooperative 3D virtual worlds feasible. Collaborative applications within a shared virtual space have huge potential.

3D Document creation

Existing user interfaces are designed with a 2 dimensional interface in mind, and are usually based on the idea of "page" or "document". With the increasing uptake of 3D applications, a new metaphor is needed. This project would investigate and propose metaphors for 3D document creation, usage, and manipulation.

Modelling Users of Search Engines

Search engine and portal providers are continually trying to improve search engine usability. One way of doing this is to build a model of a users' behaviour during searching. This goes beyond knowing which pages a user visits. It is also useful to put searching into the context of the wider operating environment - for instance, a user may perform the following actions (not exhaustive):

· cut and paste text from a page and place it in another document

· bookmark a page

· switch repeatedly between the web browser and another application

This is an experimental project which would involve building an agent which analyses users' pattern of behaviour during a search.

Visualisation of documents

There are a large number of people who have to sift through very large bodies of information to identify documents that are relevant to their goals.

The aim of this project is to build a system which will support users by providing visualisations of large document spaces. This will be done by using an existing visualisation system (HyperSpace) and building a number of agents which will wander through the document collection to find 'interesting' things. Each of the agents will be fairly simple - they might classify documents by keywords, authors, document type (e.g. index pages may have a large proportion of links) and so on. Some clues to the best places to look will be emergent from the self organisation of the visualisation.

The initial application will be the web or similar

Web Personal Assistant

Work in the School has looked at ways in which additional support can be given to users searching the web. The two main approaches used have been through the use of visualisation tools and through the use of intelligent agents.

The aim of this project is to look at ways in which a collection of agents can be used to provide support for users browsing the web. For instance, agents can be provided which will, based upon a model of the user and the user's tasks, point the user towards relevant pages or filter out irrelevant ones.

Glyph Generation

A glyph is a symbol that is typically used in data visualisation to mark data points. Some systems will allow the use of a fixed set of anonymous glyphs, in other cases meaningful symbols can be used to add dimensionality (e.g. smiling or sad faces). The aim of this project is to automatically generate glyphs based upon attributes of the data. These glyphs should be 3 dimensional and will also have shape attributes, colour, texture, movement etc. An example might be to use an object's name to generate a polyhedral shape in such a way that objects with common elements to their names would have common features to their shape (i.e. all objects with "sort" in their name would have a finger sticking out of their head.

GAs for web navigation

The strategy for web navigation can be complex. The aim of this project is to investigate ways in which NNs and GAs might be used to construct intelligent browsing assistants. These assistants will learn strategies for navigation and assist the user by either automating parts of the process or by pre-fetching pages which might be useful in order to improve response time (ie. to act as an intelligent cache).

Energy usage meters for Facebook

The aim is to build a Facebook application which will provide a ‘window’ onto a person’s energy (or water ...) usage. So, a family or business could link to their energy usage (e.g. Through a smart meter) and share this through facebook—the aim being that they could see that is being used and when and then modify their behaviour. There are many interesting and challenging extensions that could be considered.

Methods for building synthetic art

There have been many approaches to building tools that create art—mostly the use of evolutionary techniques to build 3D objects. The aim of this project is to explore the various approaches and tool build a tool that will work with a user to build novel and interesting art forms.

Room 236

School of Computer Science

The University of Birmingham

Birmingham

B15 2TT

R.J.Hendley@cs.bham.ac.uk

Contact: