Appeal 2007-1066
Application 10/170,510
candidate scorer “adjusts a confidence” as set forth in the claim. The
rejection submits that the feature was well known in the art, as evidenced by
Reed at column 50, lines 22 through 27. (Answer 5.) That section of Reed
describes a pseudo-deduction module (PDM) 58 (Fig. 21), which includes a
response selection submodule 212 to determine whether there is sufficient
confidence that the highest scoring response category is the “best” response
category from among the possible response categories.
Appellants argue that selection submodule 212 does not adjust the
confidence. According to Appellants, Reed provides no teaching that the
confidence is adjusted but simply outputs an indication that a reliable answer
could not be determined if the highest score is not sufficient. (Br. 9.) The
Examiner, in turn, finds that Reed at column 50, lines 22 through 27
suggests that a confidence can be measured by, or based on, the associated
higher or highest score. The combination of Su and Reed would thus have
suggested providing a confidence measured by, or based on, a score
mechanism, so that a score adjustment would be associated with the
adjustment of the confidence (i.e, a confidence score). (Answer 10-11.)
Appellants respond, in turn, that Su teaches adjustment of parameters for a
scoring mechanism, rather than an adjustment of a score. The combination
of Su and Reed thus may teach adjusting parameters for determining a
confidence, but does not teach adjusting the confidence itself. (Reply Br. 2.)
The Examiner relies also, however, on Potamianos for a teaching of
adjusting a confidence. (Answer 6-7.) Appellants respond that Potamianos
does not teach adjusting a confidence based on system intent. According to
Appellants, Potamianos teaches using the confidence score to obtain a single
3
Page: Previous 1 2 3 4 5 6 7 Next
Last modified: September 9, 2013