Ex Parte Sakai et al

Board of Patent Appeals and Interferences

May 17, 2011

11007552 (B.P.A.I. May. 17, 2011)

UNITED STATES PATENT AND TRADEMARK OFFICE
____________

BEFORE THE BOARD OF PATENT APPEALS
AND INTERFERENCES
____________

Ex parte DAI SAKAI, MASAMI TADA, AYA MORI,
and HIROBUMI TOYOSHIMA
____________

Appeal 2011-003522
Application 11/007,552
Technology Center 2100
____________

Before LANCE LEONARD BARRY, JAY P. LUCAS, and
JOHN A. JEFFERY, Administrative Patent Judges.

JEFFERY, Administrative Patent Judge.

DECISION ON APPEAL
Appellants appeal under 35 U.S.C. § 134(a) from the Examiner’s
rejection of claims 1, 3, 5-7, 9, 10, 12, 14, 16, 18, and 20. We have
jurisdiction under 35 U.S.C. § 6(b). We affirm-in-part.

STATEMENT OF THE CASE
Appellants’ invention dynamically classifies and sorts search results
according to natural language queries. See generally Abstract. Claim 1 is
illustrative:
Appeal 2011-003522
Application 11/007,552

2
1. An information retrieval system comprising:

an input unit for entering a query in natural language;

a natural language processing unit for performing natural
language analysis on the query centered from said input unit;

a search unit for retrieving information using at least one
keyword obtained through the natural language analysis of the
query by said natural language processing unit;

a search result processing unit for analyzing information
related to the keyword obtained through the natural language
analysis of the query by said natural language processing unit
based on predefined sematic content of the information to
process the results of the information retrieval by said search
unit based on the analysis result, wherein

said search result processing unit analyzes a modifier of
the keyword included in the query using an ontology describing
semantic content to interpret a restrictive condition of the
keyword and sort the search results from said search unit based
on the restrictive condition; and

an output unit for outputting the search results processed
by said search result processing unit.

The Examiner relies on the following as evidence of unpatentability:
Omoigui US 2003/0126136 A1 July 3, 2003
Busch US 7,027,974 B1 Apr. 11, 2006
(filed Oct. 27, 2000)

Appeal 2011-003522
Application 11/007,552

3
THE REJECTION
The Examiner rejected claims 1, 3, 5-7, 9, 10, 12, 14, 16, 18, and 20
under 35 U.S.C. § 103(a) as unpatentable over Omoigui and Busch. Ans.
4-15.1

CONTENTIONS
Regarding representative claim 1, the Examiner finds that Omoigui
discloses an information retrieval system with every recited feature except
for a search result processing unit that analyzes a keyword modifier included
in a query using an ontology describing semantic content to (1) interpret a
restrictive condition of the keyword, and (2) sort the search results from the
search unit based on the restrictive condition. Ans. 4-5. The Examiner,
however, cites Busch as teaching these features in concluding the claim
would have been obvious. Ans. 5-6, 15-26.
In reaching this conclusion, the Examiner maps the recited keyword
modifier to Busch’s “concepts” which are said to include the individual
words of sentences that are analyzed by Busch’s ontological parser to create
predicate structures. Ans. 22. The Examiner also maps the recited
“restrictive condition” to two features in Busch: (1) the created predicate
structure, or, alternatively, (2) filtering sentences based on certain
conditions. Ans. 17.

1 Throughout this opinion, we refer to (1) the Appeal Brief filed June 28,
2010; (2) the Examiner’s Answer mailed October 7, 2010; and (3) the Reply
Brief filed December 7, 2010.
Appeal 2011-003522
Application 11/007,552

4
Appellants argue that Busch fails to cure the noted deficiencies of
Omoigui as the Examiner contends. First, Appellants argue that Busch’s
“concepts” are not keyword modifiers as claimed, let alone are such
modifiers analyzed using an ontology describing semantic content as
claimed. App. Br. 11-13; Reply Br. 7-9. Appellants further contend that
Busch does not use this analysis to interpret a restrictive condition of the
keyword as claimed since (1) the predicate structure (which the Examiner
maps in part to the “restrictive condition”) includes the keyword, and (2) the
filters are not generated based on applying Busch’s ontology, but rather
constitute separate processing steps. App. Br. 13-15; Reply Br. 9-11.
Appellants add that Busch does not sort the search the results from the
search unit based on the restrictive condition as claimed since Busch’s
predicate structures are the results, and therefore not used as a basis to sort
search results. App. Br. 15-19; Reply Br. 11-12. Lastly, Appellants contend
that the Examiner’s asserted reason to combine the cited references is
unsubstantiated and therefore improper. App. Br. 19-22; Reply Br. 15-16.
Regarding representative claim 3, Appellants argue that the cited prior
art does not teach or suggest an acquired lower category of a keyword, let
alone that this lower category is used to classify the search results as
claimed. App. Br. 22-23. Appellants make similar arguments regarding the
entered item of claim 5. App. Br. 23-24. The issues before us, then, are as
follows:

ISSUES
1. Under § 103, has the Examiner erred by finding that Omoigui and
Busch collectively would have taught or suggested:
Appeal 2011-003522
Application 11/007,552

5
(1) a search result processing unit that analyzes a keyword modifier
included in a query using an ontology describing semantic content to (a)
interpret a restrictive condition of the keyword, and (b) sort the search
results from the search unit based on the restrictive condition as recited in
claim 1?
(2) a search result processing unit acquires a lower category of the
keyword or entered item to classify the search results from the search unit by
the acquired category as recited in claims 3 and 5?

2. Is the Examiner’s reason to combine the teachings of these
references supported by articulated reasoning with some rational
underpinning to justify the Examiner’s obviousness conclusion?

FINDINGS OF FACT (FF)
1. The Examiner’s findings on Page 4 of the Answer regarding
Omoigui’s disclosure pertaining to the (1) input unit; (2) natural language
processing unit; (3) search unit; (4) search result processing unit; and (5)
output unit recited in claim 1 are undisputed.
2. Busch’s ontological parser converts natural-language text into a
predicate-argument format that can be used by various applications,
including search engines. Busch’s ontological parser comprises two main
functional elements: (1) sentence lexer 100 that converts received sentences
into sequences of ontological entities that are tagged with part-of-speech
information, and (2) parser 200 that converts these sequences into predicate
Appeal 2011-003522
Application 11/007,552

6
structures by analyzing the sentences’ grammatical structure and applying
rules. Busch, Abstract; col. 6, l. 49 – col. 7, l. 30; col. 8, l. 45 – col. 9, l. 10;
Figs. 1-2.
3. Busch’s sentence lexer 100 comprises lexer 130 that (1) receives
sentences, and (2) passes each individual word to ontology 140. If the word
exists in the ontology 140, it is returned as an ontological entity; if not, the
word is tagged with default assumptions about its ontological status. Busch,
col. 7, ll. 15-24; Fig. 1.
4. After lexer 130 checks the last word in a sentence against the
contents of ontology 140, the unparsed sentence is passed to a series of lexer
filters 150, including adjective and adverb filters. Busch, col. 7, ll. 25-40;
Fig. 1.
5. The adjective filter removes lexemes representing adjective
concepts from sentences. If the adjective filter finds a noun that (1) follows
an adjective, and (2) satisfies the adjective’s restrictions, the adjective filter
will apply the adjective’s selectional features to the noun. Busch, col. 7, ll.
45-59.
6. The adverb filter removes lexemes containing adverb concepts
from sentences. Since adverbs do not change the meaning of their
accompanying verbs, adverbs are removed to simplify parsing. Busch, col.
8, ll. 16-20.
7. Busch’s parser 200 comprises a parser 230 that parses sentences
210 consisting of ontological entities received from lexer 100 into parse
trees representing relationships between concepts in a sentence. Parse tree
converter 240 converts the parse trees into predicates that are then filtered
based on various rules. Busch, col. 8, ll. 45-67; col. 9, ll. 41-58; Fig. 2.
Appeal 2011-003522
Application 11/007,552

7
8. Busch provides an example of parsing the sentence “The octopus
has a heart” to produce the predicate “have.” A similar
predicate is obtained by parsing a user’s search query “Do octopuses have
hearts?” in connection with a search engine. Thus, the parser enables
information retrieval using natural language. Busch, col. 16, l. 31 – col. 17,
l. 21.
9. Search engines incorporating Busch’s parser can easily substitute
words corresponding to different levels of abstraction into the arguments of
a predicate, thus broadening the search. Busch, col. 5, ll. 10-14.
10. Busch notes that the term “concept” means an abstract formal
representation of meaning which corresponds to multiple generic or specific
words or phrases, or the meanings of entire sentences. Concepts may
represent the meanings of (1) individual words or phrases, or (2) entire
sentences. Busch, col. 6, ll. 21-25.
11. The search result processing unit of the present invention
analyzes a modifier (word(s) and phrase(s)) of a keyword included in a
query to interpret a restrictive condition and sort the search results based on
this condition. An example of this technique is shown in Figure 3 for the
query “I want red-framed glasses.” In this example, Appellants’ system
interprets the term “red-framed” as a modifier of the keyword “glasses” to
interpret a restrictive condition as a basis for sorting the search results.
Spec. ¶¶ 0011; Figs. 3, 6A.

Appeal 2011-003522
Application 11/007,552

8
ANALYSIS
Claims 1, 6, 7, 10, 14, and 18
At the outset, since the Examiner’s findings on Page 4 of the Answer
regarding Omoigui’s disclosure are undisputed (FF 1), we therefore limit our
discussion to the disputed aspects of the Examiner’s rejection, namely (1)
the Examiner’s reliance on Busch to cure Omoigui’s noted deficiencies, and
(2) whether the references are reasonably combinable as the Examiner
proposes. On this record, we find no error in the Examiner’s obviousness
rejection of representative claim 1.
As Appellants indicate (Reply Br. 7), claim 1 requires a query with (1)
a keyword, and (2) a keyword modifier. The Examiner maps the recited
keyword modifier to Busch’s “concepts” which are said to include the
individual words of sentences that are analyzed by Busch’s ontological
parser to create predicate structures. Ans. 22. Although Appellants argue
that these “concepts” cannot be keyword modifiers (Reply Br. 7-9), we
disagree. Busch indicates that a “concept” represents meanings of (1)
individual words or phrases, or (2) entire sentences. FF 10. Notably, these
representations can be adjectives and adverbs—terms that modify their
respective nouns and verbs. FF 5-6.2 That Busch uses the term “adjective
concepts” and “adverb concepts” is telling in this regard. Id. Moreover,
these terms would be included in the text input to Busch’s sentence lexer (FF
3)—input text that could include a search query. See FF 8.

2 See also FF 11 (citing an example from Appellants’ disclosure that
analyzes the adjective “red-framed” that modifies the keyword “glasses” to
interpret a restrictive condition).
Appeal 2011-003522
Application 11/007,552

9
Notably, these adjective and adverb concepts (i.e., keyword modifiers)
are used as a basis for filtering which at least suggests that these modifiers
are analyzed to interpret corresponding “restrictive conditions” of the
associated keywords, namely the conditions dictating filtering associated
with the identified adjective and adverb concepts. See FF 4-6. We therefore
find no error in the Examiner’s alternative position (Ans. 17) that Busch’s
filtering functionality at least suggests the recited “restrictive condition” of
the keyword.
Nor are we persuaded of error in the Examiner’s position (Ans. 17-18,
25-26) that search results would be sorted based, at least in part, on the
restrictive condition. After Busch’s sentence lexer applies the adjective and
adverb filters noted above, Busch’s parser parses resulting sentences to
ultimately determine an associated predicate structure. FF 2-7. These
predicates and associated arguments can then be used by search engines to
not only facilitate natural-language searching, but also broaden searches by
substituting different levels of abstraction into the predicate’s arguments. FF
2, 8-9. Combining this functionality with Omoigui’s natural-language
information retrieval system would predictably yield an information retrieval
system that would sort the search results based, at least in part, on the
restrictive condition as claimed. That is, search results obtained in
Omoigui’s system via a particular predicate-based query as taught by Busch
would be based, at least in part, on the restrictive condition determined in
connection with the query. Since these search results would be sorted in
some fashion for display (e.g., by how they appear on a display, relevance,
etc.), it follows that this sorting would likewise be based, at least in part, on
Appeal 2011-003522
Application 11/007,552

10
the restrictive condition determined in connection with the associated query
that generated the results. In short, we find no error in the Examiner’s
position based on the collective teachings of Omoigui and Busch.
Nor are we persuaded of error in the Examiner’s combining the
references in concluding that claim 1 would have been obvious. In this
regard, the Examiner has articulated reasoning with some rational
underpinning to justify the Examiner’s obviousness conclusion, namely that
providing Busch’s parsing technique in Omoigui’s system would yield a
faster method of finding search results by using an ontology parser. Ans. 26.
We see no error in this reasoning, for Busch’s parser strips natural-language
queries of certain words to obtain a more compact predicate structure that is
used for searching. FF 2-9. Skilled artisans would therefore recognize that
resulting search queries based on these condensed predicates would be
processed more efficiently (and therefore faster) since, among other things,
there are fewer search terms than the original query. Although Appellants
contend that the Examiner failed to substantiate this rationale by, among
other things, comparing Omoigui’s execution before and after the
modification (Reply Br. 16), we find no such analysis is required where, as
here, skilled artisans would have recognized enhancements obtained by
combining the cited references from the teachings and suggestions of the
references themselves in light of the knowledge of ordinarily skilled artisans.
Nevertheless, we find that enhancing Omoigui’s search system with Busch’s
parsing technique to be nothing more than the predictable use of prior art
elements according to their established functions—an obvious improvement.
See KSR Int’l Co. v. Teleflex Inc., 550 U.S. 398, 417 (2007).
Appeal 2011-003522
Application 11/007,552

11
We are therefore not persuaded that the Examiner erred in rejecting
representative claim 1, and claims 6, 7, 10, 14, and 18 not separately argued
with particularity.

Claims 3, 5, 9, 12, 16, and 20
We will not, however, sustain the Examiner’s rejection of claims 3
and 5 which recite, in pertinent part, a search result processing unit that
acquires a lower category of the keyword or entered item to classify the
search results from the search unit by the acquired category. The Examiner
has not explained—nor can we find—how the cited passages in Omoigui
teach or suggest that the search result processing unit acquires a lower
category of a keyword or an entered item, let alone that this lower category
is used to classify the search results as claimed. The term “lower” is a key
qualifier in this regard, for it implies that it is relative to another higher
category of the keyword such as that indicated in steps S904-S906 of Figure
9 of the present disclosure. But apart from merely quoting the claim
language and citing corresponding passages from Omoigui (Ans. 6-8), the
Examiner simply does not explain how these passages meet the recited
limitations, nor will we speculate in this regard here in the first instance on
appeal. That the Examiner failed to respond to Appellants’ arguments in
connection with claims 3 and 5 as Appellants indicate (Reply Br. 16-17)
only further undermines the basis for the Examiner’s position.
We are therefore persuaded that the Examiner erred in rejecting claim
3 and claims 5, 9, 12, 16, and 20 which recite commensurate limitations.

Appeal 2011-003522
Application 11/007,552

12
CONCLUSION
Under § 103, the Examiner did not err in rejecting claims 1, 6, 7, 10,
14, and 18, but erred in rejecting claims 3, 5, 9, 12, 16, and 20.

ORDER
The Examiner’s decision rejecting claims 1, 3, 5-7, 9, 10, 12, 14, 16,
18, and 20 is affirmed-in-part.

No time period for taking any subsequent action in connection with
this appeal may be extended under 37 C.F.R. § 1.136(a)(1)(iv).

AFFIRMED-IN-PART

pgc