On Friday (23.01) I have attended a meeting during which I met the team from Insight Research Centre which starts working on a project that involves using Flickr image dataset to analyse images from newspapers. The dataset will be used differently and for a different purpose but the discussion that the team had about it, helped me to better understand the data set itself and the great potential of uses.
The second meeting I have attended on Monday (26.01) involved the whole team from Computing and Communications that is working on that project. This meeting did not provide me with more information about my project but I found it very interesting and beneficial. I’ve learnt about different technologies such as Wayback Machine, DataSift, the approach that is taken to a big, real project such as this one and the importance of annotation for retrieval from archives such as RTE libraries.
I have also got access to the text files for Flickr dataset, to see the structure and type of information that is available for each picture. This gives me and idea what the tags, which I am mostly interested in, contain and things that I will have to eliminate (e.g. camera info) to provide most appropriate tag suggestions. The computed image features are promised to be made available this month.