There are some ways to measure the relevance between an image and the snippet, e.g. distance from the image to the snippet, and lexical similarity between the image description and the snippet, just to name a few.
At the time being, we only select an image from the same document from which the text snippet is extracted. Selecting an illustrating image from a set of annotated images for the text snippet remains our future work.
We know this is the very first step, but we are working to improve it incrementally.
J.H. Huang
On Dec 26, 2007 7:06 AM, Ned Scott ned@nedscott.com wrote:
This is pretty slick. How does the search decide which image to use for the search-result-snippits? Or rather, can you give us some more details on the technical side of the search? -- Ned Scott
On Dec 24, 2007, at 8:50 PM, J.H. Huang wrote:
Hi folks, We have just launched a new Wikipedia-based search engine - http://www.seariki.com
The idea is to combine searching and browsing effectively. For example, we find paths from a category to top categories, and hopefully this makes navigating through category hierarchy easier. At this time, only English and Chinese are supported, but we are working on indexing more languages.
Any comments will be appreciated.
Thanks and happy holiday. _______________________________________________ WikiEN-l mailing list WikiEN-l@lists.wikimedia.org To unsubscribe from this mailing list, visit: http://lists.wikimedia.org/mailman/listinfo/wikien-l
WikiEN-l mailing list WikiEN-l@lists.wikimedia.org To unsubscribe from this mailing list, visit: http://lists.wikimedia.org/mailman/listinfo/wikien-l