I am creating a database that references images. Keywords and captions will be part of the metadata. These captions and keywords are not necessarily added to the photos with the intent of facilitating quality searches.
I would like to effectively convert the information in the captions into keywords. Obviously some of the words in the captions will not be useful (e.g. "the","he","she","of",etc.) and I would exclude them, but other words will be very useful, especially if they appear in combination with other words.
I'm trying to figure out the best way to associate these keywords and words from captions with the images. Would it be best to create a table of images (filenames, path to file, size, etc.) and a table of words, where each word has an ID, and then an associative table that references the words to the images? Or perhaps two associative tables, one to link caption words and the other to link keywords?
Is this a good approach?