Screen Snapshots

  1. Various Ways to Search and Retrieve Documents
  2. The page for retrieving documents offers a variety of way to access White House Electronic Publications. It includes a number of standing search URLs that automatically run a search when the user clicks on them.

  3. Forms-Based Search Interface
  4. A forms-based interface to boolean keyword search is not only easier to use, but also makes it impossible for users to mispecify searches syntactically, if not semantically.

  5. Search Results
  6. After running a successful search, the user sees this page of results, which are generated on the fly from document records in a persistent-object database accessible to the server.

  7. Viewing a Document
  8. Clicking on a title of brings up the document.

  9. Specifying a Rule Learning Experiment
  10. This page allows the user learn if-then rules about international conflicts by selecting a collection of cases and specifying a control matrix to guide the rule learning experiment. The control matrix tells the system which independent variables to examine as it looks for rules that predict dependent variable values.

  11. Several Rules Learned over the Web
  12. The rule learner first builds a decision tree and then converts the branches of the tree into a series of rules. This page shows several rules found during the experiment specified in the previous screen.

  13. Asking a Question to a Natural Language System
  14. The user asks a question about the M.I.T. Artificial Intelligence Laboratory.

  15. Viewing the Answer to a Question
  16. After parsing and representing the question, the START system finds the answer and returns it on a Web page.