Ex parte WHITMAN (2003-1404) Paper 23
Application No. 09/532,230 Page 8
search queries in a data structure for subsequent look-up, wherein the second
component selects a search query from the log based at least in-part on a
usefulness of the search query as reflected by the post-search actions performed by
users who submitted that query,", in the context of the other claim limitations. In
this regard, the search terms ultimately selected in Bowman for inclusion in the
correlation table are not "search queries from the log," but rather are individual
search terms extracted from multiple-term search queries in the log.
Appellants wish to withdraw their remarks made in the paragraph
beginning two lines from the bottom of page 7 of the Supplemental Appeal Brief
[Paper 12], to the extent such remarks may inaccurately characterize Bowman.
The remarks on page 7 of Paper 12 are inaccurate, but they do not appear to have affected
the examiner's reasoning and consequently appear to have been harmless. Withdrawal of the
remarks is acknowledged.
[34] The examiner appears to rely on Bowman at 12:60-66 for the disclosure of the second
component, the data structure based at least in part on the usefulness of the prior query:
In other embodiments, the selection process 139 combines the related terms lists
by summing the correlation scores of terms common to other related terms lists,
without deleting any terms. Another implementation might give weighted scores
for intersecting terms such that terms appearing in more than one related terms list
are weighted heavier than those terms appearing only in a single related terms list.
[35] By itself, the cited portion only refers to selecting based on correlation scores, but when read
with the disclosure from column 11, which shows the correlation scores can be based on post-
search activities, the examiner appears to be closer to the mark.
[36] As Whitman notes, however, Bowman extracts a term, not a multi-term search query.
[37] Nothing in claim 43 requires the initial query to have more than one term.
[38] Bowman, however, produces its data structure based on the correlation of terms within a search
query.
Page: Previous 1 2 3 4 5 6 7 8 9 Next
Last modified: November 3, 2007