Overview
We are looking for PhD Students interested in making search engines (and machine learning in general) fairer, more transparent and explainable.
Minimum Requirements:
* Applicants should have, or expect to achieve, at least a 2:1 Honours degree (or equivalent) or Masters degree in a relevant subject.
* Excellent programming & especially prototyping skills.
* Ability to process large datasets.
* Ability to work with external partners.
* Understanding of the research lifecycle (ability to state hypotheses, design new methods, implement prototypes, evaluate methods and interpret results, adapt methods).
* Good knowledge of text-mining, information extraction or information retrieval would be desirable.
* Prior experience of data analysis & machine learning would be advantageous.
Every day people use search engines to find information. However, the algorithms behind search engines use a multitude of features and are trained by machine learning. As a result, the algorithms are often subject to various biases – which can have a negative impact on search performance. However, Search Engines have traditionally been evaluated in terms of efficiency and performance to judge both the speed and the quality of results returned. Of growing concern is the need to measure, monitor and mitigate algorithmic biases that creep into the search engine ranking function and unduly favour certain sites and certain documents over others.
Retrievability is a higher order document centric measure, that was introduced by Azzopardi in 2008. It describes a system's influence on how the collection is represented by the entire information retrieval system and can be used to describe how difficult it is to find each document in the collection. This novel approach to evaluation has sparked interest in the research community on how to tune systems using retrievability, avoiding the need for recourse to a test collection with test queries and relevance judgements. To date, several applications of retrievability have been explored, such as for use in document pruning and findability. However, many areas remain untouched or only partially studied. It is therefore an exciting and accessible field to explore with many unanswered questions and the potential to implement retrieval techniques that enhance the efficiency and performance of systems. Thus, we are looking for students with an interest in developing fairer search systems and exploring how performance, efficiency and retrievability bias impact search behaviours and experiences.
Further Information
The project will be supervised by Dr. Leif Azzopardi, who is an Associate Professor in the Department of Computer and Information Sciences. He leads the Interactive Information Retrieval research. The candidate will join our team within the Strathclyde iSchool, in the Department of Computer and Information Sciences, and work with a team that undertakes research on developing and understanding intelligent agents in a variety of contexts, i.e., customer service agents, conversational search agents, intelligent search agents, etc. The iSchool at Strathclyde is one of the leading and largest iSchools in the UK, consisting of over 35 staff and researchers working on a variety of problems from machine learning, algorithm bias and explainability to information seeking, retrieval and behaviour to human-computer interaction and information interaction.
While there is no funding in place for opportunities marked "unfunded", there are many different options to help you fund postgraduate research. Visit funding your postgraduate research for links to government grants, research councils funding and more that could be available.
To discuss PhD projects in this area and potential funding opportunities, please contact Dr. Leif Azzopardi (leif.azzopardi@strath.ac.uk), and send a copy of your C.V., a sample of writing (a publication or thesis you have written), a sample of code (preferably in GitHub, Bitbucket, etc.) and your interest in the topic area.
Number of places: 2
To read how we process personal data, applicants can review our 'Privacy Notice for Student Applicants and Potential Applicants' on our Privacy notices' web page.
#J-18808-Ljbffr