- Updated analyzer:
- Removed Similar Story matching. So far I haven’t been able to make it return much of anything reasonable. Needs signification work.
- Simplified interface, using tabs instead of putting everything one one page.
- Added facility to load text from .doc, .docx, .rtf, and .txt files.
- Expanded list of POV indicators for completeness.
- Updated algorithm for estimating point of view (POV) using the following logic:
- If any 1st person indicators are present, then estimate 1st.
- If no 1st person indicators are present but any 2nd person indicators are, then estimate 2nd.
- If no 1st and no 2nd person indicators are present but any 3rd person indicators are, then estimate 3rd.
- If no 1st, 2nd, or 3rd person indicators are present, then estimate unknown.
- Fixed stacking bars on the statistics page to actually stack.
- Enhanced search code, but I haven’t yet added the form fields needed to take advantage of them.
- Updated statistics to use new chart library and pull summarized data from Solr instead of computing histograms in the browser.
- Updated analyzer to include Market Gauge, Magazines with Stories in the Top Similar Stories (k-nearest neighbor), and Top 100 Similar Stories.
- Updated search to include type-ahead search based on search engine terms, as well as a pane showing raw statistics for each story.
- Added a link to download raw data in CSV. Updated the original Blog posts and added to link to the analyzer and statistics pages.
- Consolidated pages to reduce menu size and keep down the clutter.
- Added this page to track changes.