- Project Ideas for undergraduate and MSc student projects
- Work with Me for PhD, Intern, RA, PostDoc opportunities
See also:
Brain Data as Cognitive Personal Informatics
- Designing an activity tracker for cognitive, rather than physical, activity.
We have many activity trackers for our physical activities, like steps, which we are used to use as a proxy for an aspect of improving our physical health. It is becoming more and more feasible to track aspects of our cognitive activity, like mental effort (see fNIRS project below) and even stress via proxies of physical signals. But do we count stress? How do we represent cognitive activity, such that one day we might use it to monitor an aspect of our mental health and wellbeing. This project is approaching this question from several angles, from design work, to prototyping, to collecting data samples of objective brain data from in-the-wild studies.
- Papers: MobileHCI23 WS Paper 2 | MobileHCI23 WS Paper 1 | CHI2023 WS Paper | CHI2022 Paper | CHI2022 SIG | HCII2020 | Pervasive Health 2019 | fNIRS2018 Poster | CHI2018 WS Paper
- Ethics Papers: FAccT 2022 | CHIWork22
DigiTOP Project
- EPSRC project investigating mental workload in digital manufacturing.
This EPSRC funded work ( EP/R032718/1 ) will investigate new recommendations for human factors and human factors regulations in digital manufacturing, given that the work is shifting from hard physical work to complex cognitive work. Led by Prof Sarah Sharples (UoN), we will collabroate with Cranfield, Loughborough, and Bristol Robotics Lab, and work with partners including Jaguar Landrover and BAE Systems. My work package will build on top of our prior work using fNIRS to estimate Mental Workload levels, and combine it with other techniques from our project partners, to estimate mental workload and other cognitive activities related to digital manufacturing.
- Project Website: http://digitop.ac.uk | EPSRC Link
- Papers: HFEMSI (2023) | Workplace 4.0 (2022) | CHI2022 | IJHCS (2021) | Pervasive Health 2019 | fNIRS 2018 poster
- Contravision Videos: Videos
Brain+ Project
- European Funding to support brain health for people with Alzheimer's.
This project is led by Brain+, a Danish SME focused on mobile apps for brain health. After receiving european innovation funding, this project has developed a new app that may detect early signs of memory problems associated with Alzheimer's and dementia. Working in collaboration with Oxford and Aarhus, Max is leading a work package, engaging directly with people living with dimentia (Nottingham's Institute of Mental Health) in PPI workshops, and studying brain training with fNIRS (Computer Science).
- Project Website: Brain+
- Papers: IJHCS 2024 | JMIR Serious Games (2022) | HCII2021 | HCII2020 | MIND2019 | CHI2019 Workshop Poster
#Scanners - BCI Movie
- Exploring neuro-responsive cinematic techniques.
#Scanners is a new interactive visual arts installation that bridges the gap between digital arts and neuroscience. We have created an experience that uses wireless brain scanners that allows the user to manipulate a digital art installation. Narratives and layers can be built that are all governed by the users concentration and meditation levels. Edit points can be created by monitoring the users blinking. The audience can project their feelings onto the film that they are seeing, the film they watch will have a series of overlapping structures that they can interact with and/or disregard.
Our new movie - The Moment (see imdb) - will be released in June 2018, and previewed at CHI2018.
- Papers: IJHCI (2021) | Book Chapter | NIME2019 | CHI2019** | CHI2019 LBW | fNIRS2018 Poster | CHI2018 Demo | CHI2016* | CHI Preview | C&C Demo | British HCI Position Paper
- Media: BBC News | BBC Click (starts 6:45 in) | BBC Technology | Sky News | MIT Technology Review | Futurism | Tech Explore | Daily Mail
* CHI2016 Best Art Paper Award, ** CHI2019 Honorable Mention
Measuring Mental Workload with fNIRS
- Visualizing and measuring brain signal changes during tasks like search
There's a lot of cogsci work relating to information seeking and sensemaking, including cognitive load theory. This new project aims to integrate measurements of brain signals other indicators of information seeking, including: mouse interaction data, log data, and utterances. Our initial work used the Emotive EEG sensor, but our ongoing work is using an fNIRS
- fNIRS: Neuroergonomics (2023) | IJHCS 2021b | IJHCS 2021a | NEC2021 (Poster) | HTTF2019 | TOCHI (2018) | CHI2018 Late-Breaking | CHI2016 - Usability Study | CHI2015 - Artefact Effects | CHI2014 - Speech Artefacts | EuroHCIR2013 | EHF2014.
- Earlier EEG work: euroHCIR2011 | UIIR 2009
Confusing Search & Search Literacy
- Improving search literacy, so searchers can solve confusing problems by themselves
Many people encounter problems, like tech problems, which they find hard to solve - they dont really understand the technology, and so dont know what words to search with, and cant tell if the results apply to their situation. This happens with domain-novices, but also with domain-experts who lack specific points of knowledge to bring answers together. Together with Prof Mike Twidale (Illinois), we are studying social Q&A for tech problems, and evaluating novel Search User Interface design ideas, where search engines help users to develop their search literacy in order to find the answers for themselves.
- IP&M (2019) | CHIIR2019* | CHIIR2017 - Tetris Model | Complex Task WS | SAL2016 | CHIIR2016
*Honourable Mention
A Comprehensive Model of Web Session Boundaries
- Finding better ways to divide web history into sessions than naive time-outs
Search Engines are increasingly interested in using your recent, rather than global, history to provide more relevant results. So 'Java' results are relevant to your current search on coffee, even though your history is full of programming queries. Methods of determining when sessions start and stop is an ongoing challenge, with most logs being naively divided by large gaps; often 25.5 minutes. Our work is to build, and then practical, a comprehensive model of factors that divide, and do not divide, sessions.
- Early Work-in-Progress Paper: EuroHCIR2013 | Full Paper: IIiX2014
Social Media Behaviour
- Investigating appropriate retrieval techniques for fast-paced social media content
Information Retrieval is a long well studied field, where the notions of relevance have been clearly defined in a systems perspective, but unclearly defined from a human, task aware perspective. With social media the problem is even harder - content is short, ephemeral, casual, temporally relevant, and in high-volume.
Our projects have investigated a) the usefulness of tweets, b) tweets about mental health, and c) favoriting behaviour on Twitter.
- ICWSM 2011 - Tweet Usefulness | ICWSM 2014 - Twitter Favoriting | IIiX2014 - tweets about depression
- Media Coverage: Slate | Washington Post | Buzzfeed | Dailydot | Brafton
A Domain-Agnostic measure of Sensemaking/Learning
- Using Bloom's taxonomy to measure knowledge gain during undirected web search.
Sensemaking and Learning is a hard thing to measure, especially when people are searching to uncontrolled resources like the web. You need to bound the learning to use a quiz, fact counting is distorted by useless facts, and topic breadth/depth judgements require domain knowledge. Instead, we are developing a measure that focuses on depth of learning, while remaining domain-agnostic, looking for signs of higher levels of Bloom's taxonomy. This measures more than just recall, but whether participants are joining information together and critiquing the reliability of the results.
- About the measure: JASIST 2013
- Early work: CHI2011 WiP | CHI2012 WiP
Casual-Leisure Information Behaviour
- A new model of casual searching behaviour
Off the back of the social-media-ethnographic investigation into peoples search experiences, as discussed freely by Twitter users (see project below), David Elsweiler (Erlangen University) and I have combined our recent findings to produce the first version of a new model of casual-leisure search.
Unlike traditional information science models of information seeking behaviour, our recent findings focus on the casual, for-fun, entertainment value that people get from searching not from what they necessarily find. These experiences are often lengthy and dependent on being able to search, not being able to find. Further, this kind of search is becoming increasingly important as technology becomes more pervasive within our casual non-work lives.
*Award: Outstanding Author Contribution
Information vs Interaction
- Separating / measuring the benefits of information & interaction in search interfaces
This new project is focusing on understanding the importance of good metadata and good interaction. Many of the successful advances in search user interface design come with great interaction and richer better metadata. Faceted Browsing is a powerful interaction for browsing datasets, but requires carefully produced metadata. Which did the user benefit from?
Understanding these elements are important, because sometimes datasets and records are fixed - or so large that it is difficult or too expensive to produce new metadata for. So can we just improve the interaction? If we provide faceted-like interaction over simple metadata - do the searchers beenfit?
- This project will produce guidance for search user interface designers about the value of investing in information, and interaction.
- CHI2009 Sensemaking Workshop | CHI2011 Ext. Abstract | CHI2012 Ext. Abstract
Mapping Neighbourhoods with Internet Derived Data
- Using Internet Derived Data to Map Informal Boundaries of Neighbourhoods
Neighbourhoods are important local boundaries in our lives that help us to define where we live and in which communities. Yet our neighbourhoods are not formally bounded. We are working on using the data reported by people and businesses online to see how and where people self-identify as living.
This is no easy challenge. Unlike voting and postal bondaries, which are formally defined, there is disparity as to where people believe they live. Consequently, a fuzzy model is required to model these boundaries.
Sii: Search interface inspector
- Evaluate your search user interface in minutes!
Sii is designed to quickly answer: How good is your search interface?
Sii can inspect search interfaces, and even prototype designs, for how well they will support different types of search tactics. Uniquely, Sii also evaluates how useful the design is for different types of searchers, whether they know exactly what they want, or are hoping to stumble across a good bargain.
Sii is designed to be used by interface designers, usability experts, or information architects to evaluate early designs and show that they are right for the target users. It only takes a couple of hours to apply the method to each design (but you get better at it quickly) and can pick up lots of usability issues early in the design process. Much faster, cheaper, and earlier than user studies.
- JASIST 2009* | IP&M 2010 | JCDL2009 Workshop | CHI2007 Workshop | Thesis | Final Model Report
- Try it! | Guide Document
*Award: Best JASIST article 2009
Casual Information Needs
- Using Twitter to learn about real human casual information needs
Figuring out what people are searching for from the keywords they type into a keyword search box alone, is a hard task. Especially when they type in things like: Orange, Java, Mole, Bank
Turns out that people talk about the things they have or are trying to find on the internet. We've been collecting these conversations about search to get insights into real search tasks and example information needs.
mSpace
- Supporting exploratory search with faceted browsing
" I don't know much about Classical Music but i know what i like when i hear it"
This is the starting motivation for mSpace: supporting access to information by providing consistent context and domain specific preview cues of information. Though our context-rich interface, conjunctive and complex queries can be processed that cannot be easily accessed through sites like Google.
The most important thing is that we want users to explore and not search. By using Semantic Web technologies we can easily expose the relationships between information so that semantically close information can be reached through simple lightweight interactions.
Continuum
- A timeline visualisation for high-dimensional data
Continuum is designed to rise past the limitations of timeline visualisations. Available timelines struggle to represent relationships between plotted events, if not entirely unable. For example, how can we clearly and easily show which pieces of music Beethoven composed, as well as when. Also, the usual approach to scaling and showing thousands of items is to simply zoom and out make everything smaller - often creating 'the blob'. In continuum, we use time and space both to convey relationships and semantically summarise data when scale is an issue.
mSpace Mobile
- Taking faceted browsing on the move.
mSpace mobile is a Mobile Application version of mSpace, designed to run on PDAs. Using .NET technology, information is transmitted wirelessly between the PDA and our server, allowing the user to browse semantic information about their surroundings. GPS is used to ascertain the user's position and provide extra context to the semantic information through a map; preview cues are presented as icons on this map.
mSpace Mobile uses custom designed ZedPanes to provide a focus+context interface, allowing the user to maintain context and focus upon the momentary interests as they go.
Tweetwahoo!
- Tweets about your search results
We think social search got it wrong so far. Tweets about keyword searches arent always useful. Tweets about links are very popular. TweetWahoo! shows you tweets about your search results to help guide you in your search. This search prototype shows you how much websites and individual results are talked about on the web at the moment.
- The number of tweets about the domain provide social insights into the popularity or influence of the website as a whole.
- Tweets about individual pages tell you more about why you should read it.
In the example picture to the right, I'd start with the 2nd result - wouldnt you?
Tweempact
- watch your tweets spread globally
Authority in twitter is pretty undefined. Some people have approved accounts, but the power of Twitter lies in being able to share anything thats happening anywhere in the world - and so an authority on an event may be Joe Average, who saw it happen.
We also like to see where information has come from and where it goes.
Tweempact shows you the spread of information, who ReTweeted it, and how far it reached. This is an early prototype and currently needs its authentication moving to OAuth. Current research is focusing on tracking implicit RTing of information too.