Overview
Topic status: We're looking for students to study this topic.
Project Summary
The project requires the development of a new index structure for a search engine. The index will support search of text, images, and other types of multimedia/data objects. This is a challenging programming project that will appeal to students who just love programming. The development will be in C++ and it requires the utilisation of parallel processing threads. Good programming skills are a pre-requisite.
The software will be part of ongoing research into Content Based Image Retrieval and Text Retrieval. Example applications are for instance, a Google search over text, a Google Images search over images (by giving an image example instead of keywords as the search argument), or searching “spoken documents” (such as recorded speech or video soundtrack) by a spoken query rather than by typing text. Of course the project offered here will address one component of that system and will not address the entire system development. The program will form part of an Open Source release of a search engine so the student/s have an opportunity to contribute to the creation of a useful resource that may be used freely by others in their applications and/or research projects.
Expected outcomes, applications and/or benefits
The project will implement a solution to the open problem of finding top-k nearest neighbour efficiently in extremely large collections. This is an essential part of any signature based search engine and there are many applications that require an efficient solution in order to scale up.
The solution will form part of the TopSig search engine, developed at QUT, and which we have released as Open Source recently.
The main benefit to the student is the opportunity to work on a cutting edge research problem as a member of a team, and get invaluable experience in multicore multithreaded programming in C++. It can lead to follow up with an Honours, Master, or PhD program, in Information Retrieval and Search Engine Technology.
Required student skills/experience
Good programming skills, good grasp of parallel processing / multithreading, and experience with C/C++ or Java. The programs will be written in C/C++.
- Study level
- Vacation research experience scholarship
- Supervisors
- QUT
- Organisational unit
Science and Engineering Faculty
- Research area
- Keywords
- search engine, information retrieval
- Contact
- Contact the supervisor for more information.
Associate Professor Shlomo Geva
