The thing you may be running into is that the Reader looks for meta-data in RTF & Word files in a specific place, and if it's not properly encoded in the file it just defaults to using the file name.
Since you have a fairly large set of files, you might want to look at
libprs500, it's got some nice tools to manage metadata for you.
It might even entice you to try converting your files to LRF format. The Reader handles that format a good deal more agilely, and the files are something like half the size.