THE TWELFTH IN A SERIES COVERING RESEARCH IN AUTOMATIC STORAGE AND RETRIEVAL, THIS REPORT IS DIVIDED INTO THREE PARTS TITLED EVALUATION, CLUSTER SEARCHING, AND USER FEEDBACK METHODS, RESPECTIVELY. THE FIRST PART, EVALUATION, CONTAINS A COMPLETE SUMMARY OF THE RETRIEVAL RESULTS DERIVED FROM SOME SIXTY DIFFERENT TEXT ANALYSIS EXPERIMENTS. IN EACH CASE, SIGNIFICANCE COMPUTATIONS ARE INCLUDED WHICH CAN BE USED TO ASSESS THE STATISTICAL SIGNIFICANCE OF THE REPORTED RETRIEVAL RESULTS. CONCLUSIONS ARE DRAWN FROM THE EXPERIMENTAL RESULTS WHICH MAY AFFECT THE DECISION OF FUTURE FULLY-AUTOMATIC INFORMATION RETRIEVAL SYSTEMS. THE SECOND AND THIRD PARTS ARE DEVOTED TO A NUMBER OF EXPERIMENTS IN REAL-TIME INFORMATION RETRIEVAL, DESIGNED TO MAKE THE USER PARTICIPATE IN THE SEARCH PROCESS. SPECIFICALLY, VARIOUS ITERATIVE SEARCH STRATEGIES ARE EVALUATED BASED ON FEEDBACK INFORMATION RETURNED TO THE SYSTEM BY THE USER, FOLLOWING AN INITIAL SEARCH OPERATION. THIS FEEDBACK INFORMATION IS THEN USED TO IMPROVE THE OUTPUT OBTAINED DURING SUBSEQUENT SEARCHES. EVALUATION RESULTS ARE PRESENTED IN PART THREE FOR MANY KINDS OF USER FEEDBACK STRATEGIES. THE THIRTEENTH REPORT IN THE SERIES IS LI 000 212. THIS DOCUMENT IS AVAILABLE AS PB-176-536 FROM THE CLEARINGHOUS FOR FEDERAL SCIENTIFIC AND TECHNICAL INFORMATION, SPRINGFIELD, VIRGINIA 22151, $3.00 FOR HARD COPY, $0.65 FOR MICROFICHE, 346 PAGES. (AUTHOR/CM)