In recent years, eye-trackers are getting affordable and becoming available to mass market (similar to webcams or fingerprint readers which used to be distinct and expensive devices). In the near future it is possible that many personal computers will be equipped with gaze tracking mechanism for interaction and analyzing user’s gaze information with web pages. WeST has already developed an award winning Web browser, which supports full-featured web browsing by eye movements . Furthermore, eye tracking may become another mean of interaction of ordinary people. They can benefit by using such device as a mean of additional control of computer and also gain personalized feedback adapt to their behavior.
More specifically, eye tracking can add new features to the Web, since the user’s gaze path and hence his attention can be measured (content users look at can be identified), web pages can be adapted in real time and make Web browsing more personalized and appealing. Currently, the most popular methods to collect user interest are high-level feedback like clicks, how much time you spent on a website, search queries etc. With gaze it is possible to identify more fine level detail of which sentence, paragraph user really paid attention.
In this project, your aim would be to establish a platform, which collects the Web interaction data with user’s eye movements (e.g., reading behavior), perform analysis on this data, and showcase its impact with one use-case of Web interaction.
You can imagine the user profile as a text index consisting of keywords user read more often while Web browsing. It could be short-term profile (based on a session) or based on long term browsing behavior. We will choose one of the following use-cases to showcase the usability of gaze-based user profiling:
- Ad placement: More often you see website banners or Facebook ads based on what you have searched or clicked. Can we recommend more relevant adds based on gaze-based profiles?
- Summarization: Imagine reading news or research paper online, what if the personalized service provides you a summary (or highlight) of most relevant text based on personal interest (or based on gaze-profiles of all the users who have read the document).
- Intelligent Text predictions: Usually the autosuggestions are based on language statistics or your typing history. Whether the text profile based on user’s reading behavior would provide better suggestions for end-user?
- Personalized multimedia retrieval: Reranking the Web/image search results.
- Any other use-case that you think can benefit from gaze data (we will discuss the options during the first weeks of project)
Teams of 4-8 students would usually work together on a task.
We will have a first informative meeting about the Praktikum on
Wednesday, 2 August 2017 at 17.00 in room B 017 (PDF)
Students interested in joining the project are invited to come to the introduction presentation, where we would discuss different use case scenarios.
The kick-off meeting will be on 24th of October, room and hour will be announced soon.
Questions and registration
 MAMEM project: http://www.mamem.eu