I saw recently an event at an university in Romania (Universitatea Politehnica Bucuresti) that aims to help students to choose the subject for their degree thesis. At this event companies are invited to present themes in front of students. You will find below a short list of themes related to our industry:
1. Automatic website classification
Possible categories: e-commerce, company website, news/blog, other.
2. Detecting website structure (and representing as a tree)
E.g. The first level of an online store contains main categories, second level sub categories and n level product page. The entire website can be represented as a tree.
3. Logo detection on internet
When detecting logos on a website page there are multiple issues that might occur. For example: many logos in same image, scaled logos.
Please let us know if you want to develop one of the above themes, and we will help you with results of our research.