U. Cetintemel, M. J. Franklin, and C. L. Giles. Flexible user profiles for large-scale data delivery. Technical Report CS-TR-4005 (UMIACS-TR-99-18), University of Maryland, 1999.  Home/Search   Document Not in Database   Summary   Related Articles

.... Description d document v d vector representation of d f d indication of user feedback for d (2f,1;1g) P user profile p i profile vector # threshold value (2 #0:0; 1:0#) # adaptability value (2 #0:0; 1:0#) Table 1: Basic Notation In the rest of this section, we describe MM in detail (see [8] for a pseudocode of MM, and refer to Table 1 for notation) 3.2. Constructing the profile When a new relevance judgment, f d , for a document, d, is received, MM first identifies the profile vector that is most similar to v d . This profile vector, if it exists, becomes the active profile ....

....threshold value, then the vector is removed from the profile. When two profile vectors are merged, the resulting vector obtains a strength value that is equal to the sum of the strengths of the merged vectors. The details of the deletion operation, along with illustrative examples, can be found in [8]. 3.5. Discussion There are two important parameters that control the way MM behaves and thus have to be set properly. One of them is the threshold parameter, # 2 #0:0; 1:0#, which is mainly used to decide whether a new document vector should be incorporated into an existing profile vector or ....

[Article contains additional citation context not shown here]

U. Cetintemel, M. J. Franklin, and C. L. Giles. Flexible user profiles for large-scale data delivery. Technical Report CS-TR-4005 (UMIACS-TR-99-18), University of Maryland, 1999.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC